From baca30feb1dbd52caa8ae89737ec3d9c41311bca Mon Sep 17 00:00:00 2001 From: Kjell Tore Guttormsen Date: Sat, 30 May 2026 21:27:06 +0200 Subject: [PATCH] =?UTF-8?q?feat(linkedin-studio):=20S14=20=E2=80=94=20jour?= =?UTF-8?q?ney=20layer=20(create/measure=20front-doors=20+=205-journey=20r?= =?UTF-8?q?outer),=20v4.1.0?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 14a's cold command-rationalization found ZERO redundancy across the 27 commands (no defensible merge/cut), so the operator reframed S14 from "merge/cut" to "add a journey layer over the kept atomics". - Add /linkedin:create + /linkedin:measure — delegate-only guided front-doors (Read/Glob/AskUserQuestion only; route to the command that owns the work) - Re-tier commands/linkedin.md into 5 journeys (Start/Create/Engage/Measure/Grow); onboarding/strategy elevated as Start/Grow front-doors; Engage = calendar+firsthour tier - 14a honesty nits: router now lists firsthour; calendar cross-links to firsthour; competitive confirmed UNGATED (the claimed 1K-gating inconsistency was unfounded) - Lockstep: EXPECT_COMMANDS 27->29, v4.0.0->4.1.0 across plugin.json / README badges / plugin+root CLAUDE.md / README / CHANGELOG; new README commands-badge lint guard - 14a deliverable corrected: multiplatform entry added (26/27 -> 27/27), header counts, competitive nit withdrawn - Independent /trekreview (2 Opus reviewers) raised 3 MAJORs (stale commands badge, router competitive self-contradiction, STATE.md count) — ALL remediated in-session; verdict ALLOW Gate: test-runner.sh 74/0/0; node --test 98/98; commands=29; v4.1.0 consistent. Additive/minor — no command removed/renamed/behavior-changed; reload registers create+measure. Co-Authored-By: Claude Opus 4.8 --- CLAUDE.md | 2 +- README.md | 6 +- .../.claude-plugin/plugin.json | 4 +- plugins/linkedin-studio/CHANGELOG.md | 20 + plugins/linkedin-studio/CLAUDE.md | 14 +- plugins/linkedin-studio/README.md | 11 +- plugins/linkedin-studio/commands/calendar.md | 6 +- plugins/linkedin-studio/commands/create.md | 66 +++ plugins/linkedin-studio/commands/linkedin.md | 122 +++--- plugins/linkedin-studio/commands/measure.md | 51 +++ .../linkedin-studio/commands/onboarding.md | 4 +- .../remediation/command-rationalization.md | 375 ++++++++++++++++++ .../docs/remediation/finish-plan.md | 15 + .../docs/remediation/journey-layer-design.md | 154 +++++++ .../docs/remediation/review.md | 222 +++++------ .../linkedin-studio/scripts/test-runner.sh | 13 +- 16 files changed, 881 insertions(+), 204 deletions(-) create mode 100644 plugins/linkedin-studio/commands/create.md create mode 100644 plugins/linkedin-studio/commands/measure.md create mode 100644 plugins/linkedin-studio/docs/remediation/command-rationalization.md create mode 100644 plugins/linkedin-studio/docs/remediation/journey-layer-design.md diff --git a/CLAUDE.md b/CLAUDE.md index 51fb854..7b5c2c2 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -9,7 +9,7 @@ plugins/ ai-psychosis/ v1.0.0 — Interaction awareness (sycophancy, reinforcement loops) config-audit/ v3.1.0 — Configuration intelligence (health, opportunities, auto-fix, whats-active) graceful-handoff/ v2.1.0 — Auto-trigger handoff via Stop hook (skill + JSON pipeline + 4-step model-aware context resolution) - linkedin-studio/ v4.0.0 — Full-spectrum LinkedIn content engine (short-form feed + long-form newsletter). **v4.0.0 is an audit-remediation release (Voyage Phase 0–3)**: a critical self-review found overclaiming, dormant capability, and structural rot, so every user-facing claim is made honest or removed, **all 11 orphaned agents are wired** (no deletions → 19 agents), a **`/linkedin:firsthour`** command is added (→ 27 commands) alongside a short-form de-AI gate + a video quality gate, `post-feedback-monitor` is promoted to Opus, the newsletter-distribution / profile-SEO / outreach surfaces are made honest, the **algorithm signals are reconciled to one sourced statement** (no model name or date; `references/algorithm-signals-reference.md` is the single source of truth), the analytics fresh-clone crash is fixed, the voice-profile leak is closed (placeholder + sentinel + gitignore), and the structure lint is rebuilt with version/count/stat-consistency guards. Breaking — reinstall/reload required for the newly-wired agents; consolidates the v3.0.0 identity break. **v3.1.0 added a cold adversarial review package (Endring 9)** to the long-form pipeline: three new headless archetypes (all Opus) — `content-reviewer` (argument integrity C1–C5, ≤8 flags), `language-reviewer` (Norwegian L1–L5, ≤10 flags), `fact-reviewer` (cold re-verification F1–F4 + pivot-risk, web search) — that re-review a frozen draft with NO drafting-session context (each refuses drafting framing as "context pollution"); new **Step 6.5 (headless-review)** in `/linkedin:newsletter` after the persona sweep, before lock (the independence layer the in-session gates can't be); standalone **`/linkedin:headless-review`** command (run in a fresh session for max isolation); **`/linkedin:pivot`** command + pivot-detection gate (>20 % word-count / >2 new sections re-opens cleared gates before lock); **per-artifact personas** (`articles.NN.personas`). Motivated by Del 4 (Security Champions pivot): the in-session editor + persona sweep shared the drafting session's framing-bias, so the shipped version was never independently re-reviewed. Pipeline 15→16 phases; 24→26 commands; 16→19 agents; additive `personas`/`pivots`/`headlessReview` state; backward-compatible, reload required for the new agents. **v3.0.0 renamed from `linkedin-thought-leadership`** (LinkedIn Thought Leadership → LinkedIn Studio): slug + agent namespace (`linkedin-studio:`) + runtime state path (`~/.claude/linkedin-studio.local.md`) all change; the `/linkedin:*` commands are unchanged (frontmatter-namespaced, slug-independent). Breaking — reinstall required; functionality byte-identical to v2.4.0. v2.0.0 consolidated surface (27→24 commands, 16→14 agents) + added `/linkedin:newsletter` orchestrator with fact-check + persona-sweep gates BEFORE lock. v2.1.0 added skeleton-gate BEFORE prose (Step 2.5 + Step 3a) + third `persona-reviewer` mode (`skjelett`); pipeline 11→13 phases. v2.2.0 hardened the longform gates (2nd production run): blocking persona hard-fails, fact-check post-cutoff web-search mandate + orthogonal-to-narrative rule, new `voice-scrubber` agent (Opus, de-AI + Norwegian-chronicle voice; gold standard = approved Norwegian editions NOT English post corpus), render+annotate operator gates (2.5/3a), edition-state reconciled with STATE.md (`edition-HANDOVER.md` deleted); agents 14→15. v2.3.0 added **Step 7.5 (visual-assets)** to `/linkedin:newsletter` — cover (+ inline figures) or carousel deck, generated (default mcp-image; external `cover-raw.png` accepted) + operator-gated via `SendUserFile` BEFORE lock so `build-linkedin.mjs` picks up `cover.png` without a post-lock re-render; pipeline 13→14 phases, new `config/image-credit-caption.template.md`, additive `visualAssets` state. v2.4.0 adds **Step 5.5 (editorial-review)** to `/linkedin:newsletter` — new `editorial-reviewer` agent (Opus) judging craft (prosa-håndverk + narrativ-arkitektur), not reader-response, ≤10 flags BLOCK/REWORK/NICE as direction, operator-gated via `SendUserFile` between fact-check (5) and persona-sweep (6), mirroring Maskinrommet skrivekontrakt §C2; motivated by Del 4 (every persona PASS yet 8 fresh editor points, ~6/8 craft/architecture blind spots). Pipeline 14→15 phases; agents 15→16; additive `editorialReview` state; doc/orchestration-only (new agent + fasit fixture + lint test the only new files). Commands unchanged (24); agents 16. Render pipeline self-hosted (OFL-1.1 fonts). + linkedin-studio/ v4.1.0 — Full-spectrum LinkedIn content engine (short-form feed + long-form newsletter). **v4.0.0 is an audit-remediation release (Voyage Phase 0–3)**: a critical self-review found overclaiming, dormant capability, and structural rot, so every user-facing claim is made honest or removed, **all 11 orphaned agents are wired** (no deletions → 19 agents), a **`/linkedin:firsthour`** command is added (→ 27 commands) alongside a short-form de-AI gate + a video quality gate, `post-feedback-monitor` is promoted to Opus, the newsletter-distribution / profile-SEO / outreach surfaces are made honest, the **algorithm signals are reconciled to one sourced statement** (no model name or date; `references/algorithm-signals-reference.md` is the single source of truth), the analytics fresh-clone crash is fixed, the voice-profile leak is closed (placeholder + sentinel + gitignore), and the structure lint is rebuilt with version/count/stat-consistency guards. Breaking — reinstall/reload required for the newly-wired agents; consolidates the v3.0.0 identity break. **v3.1.0 added a cold adversarial review package (Endring 9)** to the long-form pipeline: three new headless archetypes (all Opus) — `content-reviewer` (argument integrity C1–C5, ≤8 flags), `language-reviewer` (Norwegian L1–L5, ≤10 flags), `fact-reviewer` (cold re-verification F1–F4 + pivot-risk, web search) — that re-review a frozen draft with NO drafting-session context (each refuses drafting framing as "context pollution"); new **Step 6.5 (headless-review)** in `/linkedin:newsletter` after the persona sweep, before lock (the independence layer the in-session gates can't be); standalone **`/linkedin:headless-review`** command (run in a fresh session for max isolation); **`/linkedin:pivot`** command + pivot-detection gate (>20 % word-count / >2 new sections re-opens cleared gates before lock); **per-artifact personas** (`articles.NN.personas`). Motivated by Del 4 (Security Champions pivot): the in-session editor + persona sweep shared the drafting session's framing-bias, so the shipped version was never independently re-reviewed. Pipeline 15→16 phases; 24→26 commands; 16→19 agents; additive `personas`/`pivots`/`headlessReview` state; backward-compatible, reload required for the new agents. **v3.0.0 renamed from `linkedin-thought-leadership`** (LinkedIn Thought Leadership → LinkedIn Studio): slug + agent namespace (`linkedin-studio:`) + runtime state path (`~/.claude/linkedin-studio.local.md`) all change; the `/linkedin:*` commands are unchanged (frontmatter-namespaced, slug-independent). Breaking — reinstall required; functionality byte-identical to v2.4.0. v2.0.0 consolidated surface (27→24 commands, 16→14 agents) + added `/linkedin:newsletter` orchestrator with fact-check + persona-sweep gates BEFORE lock. v2.1.0 added skeleton-gate BEFORE prose (Step 2.5 + Step 3a) + third `persona-reviewer` mode (`skjelett`); pipeline 11→13 phases. v2.2.0 hardened the longform gates (2nd production run): blocking persona hard-fails, fact-check post-cutoff web-search mandate + orthogonal-to-narrative rule, new `voice-scrubber` agent (Opus, de-AI + Norwegian-chronicle voice; gold standard = approved Norwegian editions NOT English post corpus), render+annotate operator gates (2.5/3a), edition-state reconciled with STATE.md (`edition-HANDOVER.md` deleted); agents 14→15. v2.3.0 added **Step 7.5 (visual-assets)** to `/linkedin:newsletter` — cover (+ inline figures) or carousel deck, generated (default mcp-image; external `cover-raw.png` accepted) + operator-gated via `SendUserFile` BEFORE lock so `build-linkedin.mjs` picks up `cover.png` without a post-lock re-render; pipeline 13→14 phases, new `config/image-credit-caption.template.md`, additive `visualAssets` state. v2.4.0 adds **Step 5.5 (editorial-review)** to `/linkedin:newsletter` — new `editorial-reviewer` agent (Opus) judging craft (prosa-håndverk + narrativ-arkitektur), not reader-response, ≤10 flags BLOCK/REWORK/NICE as direction, operator-gated via `SendUserFile` between fact-check (5) and persona-sweep (6), mirroring Maskinrommet skrivekontrakt §C2; motivated by Del 4 (every persona PASS yet 8 fresh editor points, ~6/8 craft/architecture blind spots). Pipeline 14→15 phases; agents 15→16; additive `editorialReview` state; doc/orchestration-only (new agent + fasit fixture + lint test the only new files). Commands unchanged (24); agents 16. Render pipeline self-hosted (OFL-1.1 fonts). **v4.1.0 (Voyage S14)** adds a journey layer over the command surface: 14a's cold command-rationalization found zero redundancy (no merges/cuts), so two new guided front-doors (`/linkedin:create`, `/linkedin:measure`) are added and the router is re-tiered into five journeys (Start · Create · Engage · Measure · Grow) with `onboarding`/`strategy` elevated as the Start/Grow front-doors and the 27 atomic commands kept as the execution tier (27→29 commands; additive/minor, reload registers the two new commands). llm-security/ v7.7.2 — Security scanning, auditing, threat modeling. HTML report output for all 18 skill commands (render-report CLI + canonical ESM module mirrored bit-identical into the playground). v7.7.2 translated the remaining Norwegian surface text in the playground UI, the canonical renderer, the agent prompts, and the README/CLAUDE.md state sections to English. v7.7.1 stripped the playground to the catalog as the only routable surface. ms-ai-architect/ v1.15.0 — Microsoft AI architecture (Cosmo Skyberg persona) + manual KB-refresh slash command + v3 project-view (sidebar med 17 artifacts + main + import-modal overlay, v2-surface fjernet i v1.15.0) okr/ v1.0.0 — OKR guidance for Norwegian public sector diff --git a/README.md b/README.md index 751fcb2..3d7ba55 100644 --- a/README.md +++ b/README.md @@ -114,7 +114,7 @@ Key commands: `/architect`, `/architect:ros`, `/architect:security`, `/architect --- -### [LinkedIn Studio](plugins/linkedin-studio/) `v4.0.0` +### [LinkedIn Studio](plugins/linkedin-studio/) `v4.1.0` Build authentic LinkedIn authority through algorithmic understanding, strategic consistency, and AI-assisted content creation. @@ -123,9 +123,9 @@ Build authentic LinkedIn authority through algorithmic understanding, strategic - **Content engine** — Content Matrix (40+ ideas from one topic), voice training with drift detection, full ideation → publish → 48-hour monitoring → analytics - **Growth and monetization** — phase-specific guidance from 0 to 10K+ followers; topic-relevance profile optimization aligned to LinkedIn's 2026 ranking model -Key commands: `/linkedin:onboarding`, `/linkedin:post`, `/linkedin:quick`, `/linkedin:newsletter`, `/linkedin:carousel`, `/linkedin:report` +Key commands: `/linkedin:create` + `/linkedin:measure` (journey front-doors), `/linkedin:onboarding`, `/linkedin:post`, `/linkedin:newsletter`, `/linkedin:report` -19 agents · 27 commands · 6 skills · 9 hooks · [Full documentation →](plugins/linkedin-studio/README.md) +19 agents · 29 commands (five journeys) · 6 skills · 9 hooks · [Full documentation →](plugins/linkedin-studio/README.md) --- diff --git a/plugins/linkedin-studio/.claude-plugin/plugin.json b/plugins/linkedin-studio/.claude-plugin/plugin.json index b859b4d..96c2f7c 100644 --- a/plugins/linkedin-studio/.claude-plugin/plugin.json +++ b/plugins/linkedin-studio/.claude-plugin/plugin.json @@ -1,7 +1,7 @@ { "name": "linkedin-studio", - "version": "4.0.0", - "description": "LinkedIn Studio — full-spectrum LinkedIn content engine: feed posts, carousels, video scripts, and long-form newsletter editions, with the 2026 relevance-ranking model baked in. v4.0.0 is an audit-remediation release (Voyage Phase 0–3): every user-facing claim is made honest or removed, all 11 previously-orphaned agents are wired (→ 19 agents), a `/linkedin:firsthour` post-publish command is added (→ 27 commands), the algorithm-signal claims are reconciled to one sourced statement (no unpublishable model name or date), short-form de-AI and video quality gates are added, and the structure lint is rebuilt to guard the real layout plus version/count/stat consistency. Breaking: the newly-wired agents register only on reinstall/reload, and this consolidates the v3.0.0 identity break (slug, agent namespace `linkedin-studio:`, state-file path `~/.claude/linkedin-studio.local.md`). v3.1.0 added the cold adversarial review package (`/linkedin:headless-review` + Step 6.5 + `/linkedin:pivot` + per-artifact personas); the `/linkedin:*` commands are unchanged.", + "version": "4.1.0", + "description": "LinkedIn Studio — full-spectrum LinkedIn content engine: feed posts, carousels, video scripts, and long-form newsletter editions, with the 2026 relevance-ranking model baked in. v4.0.0 is an audit-remediation release (Voyage Phase 0–3): every user-facing claim is made honest or removed, all 11 previously-orphaned agents are wired (→ 19 agents), a `/linkedin:firsthour` post-publish command is added (→ 27 commands), the algorithm-signal claims are reconciled to one sourced statement (no unpublishable model name or date), short-form de-AI and video quality gates are added, and the structure lint is rebuilt to guard the real layout plus version/count/stat consistency. Breaking: the newly-wired agents register only on reinstall/reload, and this consolidates the v3.0.0 identity break (slug, agent namespace `linkedin-studio:`, state-file path `~/.claude/linkedin-studio.local.md`). v3.1.0 added the cold adversarial review package (`/linkedin:headless-review` + Step 6.5 + `/linkedin:pivot` + per-artifact personas); the `/linkedin:*` commands are unchanged. v4.1.0 adds a journey layer: two guided front-doors (`/linkedin:create`, `/linkedin:measure`) plus a router re-tiered into five journeys (Start · Create · Engage · Measure · Grow), with the 27 existing commands kept as the execution tier (→ 29 commands; additive, reload registers the two new commands).", "author": { "name": "Kjell Tore Guttormsen" }, diff --git a/plugins/linkedin-studio/CHANGELOG.md b/plugins/linkedin-studio/CHANGELOG.md index afbd9be..e04658d 100644 --- a/plugins/linkedin-studio/CHANGELOG.md +++ b/plugins/linkedin-studio/CHANGELOG.md @@ -5,6 +5,26 @@ All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). +## [4.1.0] - 2026-05-30 + +### Summary +**Journey layer over the command surface (Voyage S14).** Step 14a's cold, independent command-rationalization (`docs/remediation/command-rationalization.md`) audited all 27 commands under the same subsumption discipline as the agent-overlap study and found **zero redundancy** — the surface was over-grown in *count*, not in *duplication* (the two prior consolidations already removed the genuine overlaps). So instead of cutting, this release **organizes**: the 27 atomic commands are kept as the execution tier, and a **journey layer** is added on top. Additive and minor — no command removed, renamed, or behavior-changed; the two new commands register on reload. Design contract: `docs/remediation/journey-layer-design.md`. + +### Added +- **`/linkedin:create`** — Create front-door. One guided "what do you want to make?" entry that routes to the command owning the format (`post`/`quick`/`react`/`carousel`/`video`/`multiplatform`/`batch`/`newsletter`). Delegates only — no drafting logic of its own. +- **`/linkedin:measure`** — Measure front-door. One guided "how am I doing?" entry that routes to the right analytics command (`import`/`report`/`analyze`/`audit`/`ab-test`). Delegates only. Commands 27 → 29. + +### Changed +- **Router re-tiered into five journeys** (`commands/linkedin.md`) — Start · Create · Engage · Measure · Grow, each headed by a front-door (`onboarding`/`strategy` elevated as the Start/Grow front-doors; `create`/`measure` new; Engage is a `calendar` + `firsthour` relay), with the atomic commands nested as the execution tier. Absorbs the planned router-tiering UX step. +- **Honesty nits from 14a fixed** — the router now lists `/linkedin:firsthour` (was agent-only under Post-Publish); `calendar`'s publish-action first-hour block cross-links to `/linkedin:firsthour` for the full worked sprint plan. +- **`EXPECT_COMMANDS`** in `scripts/test-runner.sh` 27 → 29; CLAUDE.md / README rosters + counts updated in lockstep. + +### Fixed +- **14a deliverable correction** — the cold review's first pass omitted `multiplatform` (covered 26/27) and raised a `competitive` 1K-gating inconsistency that did **not** survive verification (CLAUDE.md `:64`, README `:222`, the router, and the command body all leave `competitive` ungated). Both corrected in `command-rationalization.md`; net finding unchanged (keep 27, 0 merge, 0 cut). + +### Compatibility +- **Minor / additive.** No command removed, renamed, or behavior-changed; the 27 existing commands, all 19 agents, and all state shapes are unchanged. The two new commands (`create`, `measure`) register when the plugin command set is rebuilt at session start — **reload required** to see them. + ## [4.0.0] - 2026-05-30 ### Summary diff --git a/plugins/linkedin-studio/CLAUDE.md b/plugins/linkedin-studio/CLAUDE.md index 4005fce..9d89249 100644 --- a/plugins/linkedin-studio/CLAUDE.md +++ b/plugins/linkedin-studio/CLAUDE.md @@ -1,6 +1,6 @@ -# LinkedIn Studio Plugin (v4.0.0) +# LinkedIn Studio Plugin (v4.1.0) -Full-spectrum LinkedIn content engine — short-form feed posts, carousels, video scripts, and long-form newsletter editions — with the 2026 relevance-ranking model baked in. **v4.0.0** is an **audit-remediation release (Voyage Phase 0–3)**: a critical self-review found overclaiming (tracking/analytics/review-independence the plugin couldn't deliver), dormant capability (11 agents never invoked by any command), and structural rot (a dead lint, a self-contradicting algorithm claim, an unpublishable model brand/date in user copy). The fix wires **all 11 orphaned agents** (no deletions → 19 agents), adds **`/linkedin:firsthour`** (→ 27 commands) + a short-form de-AI gate + a video quality gate, promotes `post-feedback-monitor` to Opus, makes the newsletter-distribution / profile-SEO / outreach surfaces honest, **reconciles the algorithm signals to one sourced statement** (no model name or date; `references/algorithm-signals-reference.md` is the single source of truth), fixes the analytics fresh-clone crash, closes the voice-profile leak (placeholder + sentinel + gitignore), and rebuilds the structure lint with version/count/stat/model-consistency + render-chain-propagation + `$`-safety guards (each agent's frontmatter model must match every surface declaration; no honesty pattern a command was cleaned of survives in the reference it renders from — S12; no untrusted value reaches a `String.replace` replacement string in `state-updater.mjs`, proven behaviorally + coverage-complete + self-testing — S13). Breaking — reinstall/reload required for the newly-wired agents; consolidates the v3.0.0 identity break (slug, agent namespace, state-file path). v2.0.0 consolidated the surface (27 commands → 24, 16 agents → 14) while adding the long-form `/linkedin:newsletter` orchestrator + two longform-quality gate agents (`fact-checker`, `persona-reviewer`). v2.1.0 added two gates BEFORE prose (Step 2.5 skeleton + Step 3a spine prose) + a third `persona-reviewer` mode (`skjelett`). v2.2.0 hardened the longform gates with the lessons from the second production run (Seres-serien): blocking persona hard-fails, a post-cutoff fact-check mandate, a `voice-scrubber` agent, render+annotate operator gates, and STATE.md-reconciled edition state. v2.3.0 made **visual assets an explicit pipeline phase** — Step 7.5 (visual-assets) between annotation (Step 7) and lock (Step 8): cover (+ optional inline figures) or a carousel deck, generated (default `mcp-image`, external `cover-raw.png` accepted) and operator-gated BEFORE lock so `render/build-linkedin.mjs` picks up `cover.png` at lock without a post-lock re-render. **v2.4.0** makes an **editor's craft gate an explicit pipeline phase** — new **Step 5.5 (editorial-review)** between fact-check (Step 5) and the persona sweep (Step 6): a new **`editorial-reviewer` agent** (Opus) judges **craft** (prosa-håndverk + narrativ-arkitektur), not reader-response, returning ≤10 flags (BLOCK/REWORK/NICE) as direction, **operator-gated via `SendUserFile` BEFORE the persona sweep** so the personas measure resonance instead of stumbling on craft noise. Motivated by Del 4: every persona reported PASS, yet the editor found 8 fresh points on first reading, ~6/8 of them craft/architecture blind spots no agent measured. Mirrors the Maskinrommet writing-contract §C2. Pipeline 14 → 15 phases; agents 15 → 16; additive `editorialReview` state. Doc/orchestration-only for the wiring (the new agent + its fasit fixture + lint test are the only new files); commands unchanged (24). **v3.1.0 (Endring 9)** adds an **adversarial review package** run COLD on a frozen draft — new **Step 6.5 (headless-review)** between the persona sweep (Step 6) and lock, plus a standalone **`/linkedin:headless-review`** command (run in a fresh session for maximum isolation): three new headless archetypes — **`content-reviewer`** (argument integrity), **`language-reviewer`** (Norwegian language), **`fact-reviewer`** (cold re-verification incl. claims a late pivot bolted on) — plus `persona-reviewer` in resonance + conversion modes, all with NO drafting-session context (the independence layer the in-session gates structurally cannot be). v3.1.0 also adds **`/linkedin:pivot`** (re-opens cleared gates after a late change + a >20 %/>2-section pivot-detection gate at lock) and **per-artifact personas** (`articles.NN.personas` — one or more readers configurable per edition, resolved edition-state → series file → plugin library → interactive). Pipeline 15 → 16 phases; agents 16 → 19; commands 24 → 26; additive `personas` / `pivots` / `headlessReview` state. Motivated by Del 4: the in-session editor + persona sweep shared the drafting session's framing-bias, so the version that shipped was never independently re-reviewed. +Full-spectrum LinkedIn content engine — short-form feed posts, carousels, video scripts, and long-form newsletter editions — with the 2026 relevance-ranking model baked in. **v4.0.0** is an **audit-remediation release (Voyage Phase 0–3)**: a critical self-review found overclaiming (tracking/analytics/review-independence the plugin couldn't deliver), dormant capability (11 agents never invoked by any command), and structural rot (a dead lint, a self-contradicting algorithm claim, an unpublishable model brand/date in user copy). The fix wires **all 11 orphaned agents** (no deletions → 19 agents), adds **`/linkedin:firsthour`** (→ 27 commands) + a short-form de-AI gate + a video quality gate, promotes `post-feedback-monitor` to Opus, makes the newsletter-distribution / profile-SEO / outreach surfaces honest, **reconciles the algorithm signals to one sourced statement** (no model name or date; `references/algorithm-signals-reference.md` is the single source of truth), fixes the analytics fresh-clone crash, closes the voice-profile leak (placeholder + sentinel + gitignore), and rebuilds the structure lint with version/count/stat/model-consistency + render-chain-propagation + `$`-safety guards (each agent's frontmatter model must match every surface declaration; no honesty pattern a command was cleaned of survives in the reference it renders from — S12; no untrusted value reaches a `String.replace` replacement string in `state-updater.mjs`, proven behaviorally + coverage-complete + self-testing — S13). Breaking — reinstall/reload required for the newly-wired agents; consolidates the v3.0.0 identity break (slug, agent namespace, state-file path). v2.0.0 consolidated the surface (27 commands → 24, 16 agents → 14) while adding the long-form `/linkedin:newsletter` orchestrator + two longform-quality gate agents (`fact-checker`, `persona-reviewer`). v2.1.0 added two gates BEFORE prose (Step 2.5 skeleton + Step 3a spine prose) + a third `persona-reviewer` mode (`skjelett`). v2.2.0 hardened the longform gates with the lessons from the second production run (Seres-serien): blocking persona hard-fails, a post-cutoff fact-check mandate, a `voice-scrubber` agent, render+annotate operator gates, and STATE.md-reconciled edition state. v2.3.0 made **visual assets an explicit pipeline phase** — Step 7.5 (visual-assets) between annotation (Step 7) and lock (Step 8): cover (+ optional inline figures) or a carousel deck, generated (default `mcp-image`, external `cover-raw.png` accepted) and operator-gated BEFORE lock so `render/build-linkedin.mjs` picks up `cover.png` at lock without a post-lock re-render. **v2.4.0** makes an **editor's craft gate an explicit pipeline phase** — new **Step 5.5 (editorial-review)** between fact-check (Step 5) and the persona sweep (Step 6): a new **`editorial-reviewer` agent** (Opus) judges **craft** (prosa-håndverk + narrativ-arkitektur), not reader-response, returning ≤10 flags (BLOCK/REWORK/NICE) as direction, **operator-gated via `SendUserFile` BEFORE the persona sweep** so the personas measure resonance instead of stumbling on craft noise. Motivated by Del 4: every persona reported PASS, yet the editor found 8 fresh points on first reading, ~6/8 of them craft/architecture blind spots no agent measured. Mirrors the Maskinrommet writing-contract §C2. Pipeline 14 → 15 phases; agents 15 → 16; additive `editorialReview` state. Doc/orchestration-only for the wiring (the new agent + its fasit fixture + lint test are the only new files); commands unchanged (24). **v3.1.0 (Endring 9)** adds an **adversarial review package** run COLD on a frozen draft — new **Step 6.5 (headless-review)** between the persona sweep (Step 6) and lock, plus a standalone **`/linkedin:headless-review`** command (run in a fresh session for maximum isolation): three new headless archetypes — **`content-reviewer`** (argument integrity), **`language-reviewer`** (Norwegian language), **`fact-reviewer`** (cold re-verification incl. claims a late pivot bolted on) — plus `persona-reviewer` in resonance + conversion modes, all with NO drafting-session context (the independence layer the in-session gates structurally cannot be). v3.1.0 also adds **`/linkedin:pivot`** (re-opens cleared gates after a late change + a >20 %/>2-section pivot-detection gate at lock) and **per-artifact personas** (`articles.NN.personas` — one or more readers configurable per edition, resolved edition-state → series file → plugin library → interactive). Pipeline 15 → 16 phases; agents 16 → 19; commands 24 → 26; additive `personas` / `pivots` / `headlessReview` state. Motivated by Del 4: the in-session editor + persona sweep shared the drafting session's framing-bias, so the version that shipped was never independently re-reviewed. **v4.1.0** adds a **journey layer** over the (unchanged) command surface: two guided front-doors — **`/linkedin:create`** (routes to the right creation command) and **`/linkedin:measure`** (routes to the right analytics command) — plus a router re-tiered into five journeys (Start · Create · Engage · Measure · Grow), with `onboarding`/`strategy` elevated as the Start/Grow front-doors and the 27 atomic commands kept as the execution tier (→ 29 commands). Additive: 14a's cold command-rationalization found **zero redundancy** (no merges/cuts; `docs/remediation/command-rationalization.md` + `journey-layer-design.md`), so the journey layer organizes rather than removes; reload registers the two new commands. ## Architecture @@ -32,14 +32,16 @@ Full-spectrum LinkedIn content engine — short-form feed posts, carousels, vide **Hook editing:** Edit `hooks/hooks.template.json` + `hooks/prompts/*.md`, then run `python3 hooks/scripts/compile-hooks.py`. Do not edit `hooks.json` directly. Prompts are loaded at runtime by gatekeeper scripts; the compile step is only needed when adding `type: prompt` hooks. -## Commands (27) +## Commands (29) -All content commands (post, quick, react, pipeline, first-post, video, multiplatform, carousel, newsletter) auto-copy output to clipboard via `clipboard-helper.mjs`. Interactive steps are minimized — angle, format, and post type are inferred from context, with max 2 questions per post. **v2.0.0 net change:** 5 commands removed (`templates`, `publish`, `authority`, `collab`, `speaking` — absorbed into `quick`, `calendar`, `strategy`, `outreach` respectively) + 2 commands added (`newsletter`, `outreach`) = 27 → 24. **v3.1.0** adds 2 longform companions (`headless-review`, `pivot`) = 24 → 26. **Remediation Step 16** adds `firsthour` (wiring orphan agent #11 `engagement-coach`) = 26 → 27. +All content commands (post, quick, react, pipeline, first-post, video, multiplatform, carousel, newsletter) auto-copy output to clipboard via `clipboard-helper.mjs`. Interactive steps are minimized — angle, format, and post type are inferred from context, with max 2 questions per post. **v2.0.0 net change:** 5 commands removed (`templates`, `publish`, `authority`, `collab`, `speaking` — absorbed into `quick`, `calendar`, `strategy`, `outreach` respectively) + 2 commands added (`newsletter`, `outreach`) = 27 → 24. **v3.1.0** adds 2 longform companions (`headless-review`, `pivot`) = 24 → 26. **Remediation Step 16** adds `firsthour` (wiring orphan agent #11 `engagement-coach`) = 26 → 27. **v4.1.0 (S14)** adds the two journey front-doors (`create`, `measure`) = 27 → 29. The surface is organized into **five journeys** (Start · Create · Engage · Measure · Grow); `create`/`measure` are new guided front-doors, `onboarding`/`strategy` are elevated as the Start/Grow front-doors, and the 27 atomic commands remain the execution tier (14a found zero redundancy → no merges/cuts). | Command | Purpose | |---------|---------| -| `/linkedin` | Router — status line + command menu | -| `/linkedin:onboarding` | Multi-step onboarding wizard (profile → setup → first-post) | +| `/linkedin` | Router — status line + five-journey command menu | +| `/linkedin:create` | **(v4.1) Create front-door** — guided "what to make?" → routes to the creation command that owns the format (post/quick/react/carousel/video/multiplatform/batch/newsletter); delegates only | +| `/linkedin:measure` | **(v4.1) Measure front-door** — guided "how am I doing?" → routes to the analytics command (import/report/analyze/audit/ab-test); delegates only | +| `/linkedin:onboarding` | Multi-step onboarding wizard (profile → setup → first-post); **Start-journey front-door (v4.1)** | | `/linkedin:first-post` | First-post accelerator (10 min) | | `/linkedin:setup` | Guided personalization setup | | `/linkedin:react` | URL-to-post pipeline | diff --git a/plugins/linkedin-studio/README.md b/plugins/linkedin-studio/README.md index de222eb..a2386aa 100644 --- a/plugins/linkedin-studio/README.md +++ b/plugins/linkedin-studio/README.md @@ -6,15 +6,15 @@ *AI-generated: all code produced by Claude Code through dialog-driven development. [Full disclosure →](../../README.md#ai-generated-code-disclosure)* -![Version](https://img.shields.io/badge/version-4.0.0-blue) +![Version](https://img.shields.io/badge/version-4.1.0-blue) ![Platform](https://img.shields.io/badge/platform-Claude_Code_Plugin-purple) -![Commands](https://img.shields.io/badge/commands-27-green) +![Commands](https://img.shields.io/badge/commands-29-green) ![Agents](https://img.shields.io/badge/agents-19-orange) ![Hooks](https://img.shields.io/badge/hooks-9-red) ![Reference Docs](https://img.shields.io/badge/reference_docs-25-teal) ![License](https://img.shields.io/badge/license-MIT-lightgrey) -Most experts know they *should* post on LinkedIn — and quietly don't. The blank editor wins. LinkedIn Studio turns that chore into a system: structured workflows that take you from idea to published, in your own voice, calibrated to how LinkedIn's **topic-relevance** ranking model (2026) actually distributes content. Two engines under one surface — a **feed engine** for short-form posts, carousels, and video scripts, and a **long-form engine** that runs newsletter editions and essays through a serious editorial pipeline before they ever lock. 27 commands, 19 specialized agents, 9 automated quality hooks, and a 25-document knowledge base grounded in the algorithm's real signals. +Most experts know they *should* post on LinkedIn — and quietly don't. The blank editor wins. LinkedIn Studio turns that chore into a system: structured workflows that take you from idea to published, in your own voice, calibrated to how LinkedIn's **topic-relevance** ranking model (2026) actually distributes content. Two engines under one surface — a **feed engine** for short-form posts, carousels, and video scripts, and a **long-form engine** that runs newsletter editions and essays through a serious editorial pipeline before they ever lock. 29 commands organized into five journeys (Start · Create · Engage · Measure · Grow), 19 specialized agents, 9 automated quality hooks, and a 25-document knowledge base grounded in the algorithm's real signals. This is not a shortcut. Hand the wheel to the AI and you land where everyone who did the same lands — the forgettable middle. The plugin removes the friction; the judgment, the genuine engagement, and the effort that make content worth reading remain entirely yours. @@ -183,7 +183,7 @@ The wizard handles everything: topic-relevance profile checklist, voice and user ## Commands -All 27 commands use colon notation: `/linkedin:post`, `/linkedin:quick`, etc. +All 29 commands use colon notation: `/linkedin:post`, `/linkedin:quick`, etc. The surface is organized into five journeys (Start · Create · Engage · Measure · Grow); `/linkedin:create` and `/linkedin:measure` are guided front-doors that route you to the right command when you know the journey but not the exact command. ### Onboarding @@ -197,6 +197,7 @@ All 27 commands use colon notation: `/linkedin:post`, `/linkedin:quick`, etc. | Command | Description | |---------|-------------| +| `/linkedin:create` | **Create front-door.** Guided "what do you want to make?" — routes you to the creation command that owns the format (post, quick, react, carousel, video, multiplatform, batch, or the long-form newsletter). Delegates only. | | `/linkedin:post` | Full interactive post creation with angle selection, format choice, and refinement. Best for substantial posts (1,200-1,800 characters). | | `/linkedin:quick` | 5-minute quick post using the 3-line formula. Target: 150-500 characters. Best for reactions, observations, tips, and questions. Also the single entry point for the 8 post-type templates (fill-in-the-blank structures). | | `/linkedin:newsletter` | Long-form orchestrator — newsletter edition, essay, or series article end-to-end at series quality. Multi-session 16-phase pipeline with skeleton + spine-prose gates BEFORE prose, an editorial-craft gate, fact-check, persona resonance, a cold adversarial review package, and visual assets — all BEFORE lock. | @@ -215,6 +216,7 @@ All 27 commands use colon notation: `/linkedin:post`, `/linkedin:quick`, etc. | Command | Description | |---------|-------------| +| `/linkedin:measure` | **Measure front-door.** Guided "how am I doing?" — routes you to the analytics command you need (import, report, analyze, audit, or ab-test). Delegates only. | | `/linkedin:analyze` | Analyze content performance and troubleshoot engagement issues. Diagnoses algorithm penalties, profile-content mismatches, and reach drops. | | `/linkedin:audit` | Periodic content strategy audit. Reviews top/bottom posts, topic distribution, format mix, and engagement trends. Run quarterly. | | `/linkedin:report` | Generate weekly performance report from imported analytics data. Shows key metrics, top performers, trends, and actionable alerts. | @@ -627,6 +629,7 @@ Scheduled posts are tracked in `assets/drafts/queue.json`: | Version | Date | Highlights | |---------|------|-----------| +| **4.1.0** | 2026-05-30 | **Journey layer over the command surface.** 14a's cold command-rationalization audited all 27 commands and found **zero redundancy** (over-grown in count, not duplication) → no merges/cuts. Instead, a journey layer is added: two new guided front-doors — **`/linkedin:create`** (routes to the right creation command) and **`/linkedin:measure`** (routes to the right analytics command) — and the router re-tiered into five journeys (Start · Create · Engage · Measure · Grow), with `onboarding`/`strategy` elevated as the Start/Grow front-doors and the 27 atomic commands kept as the execution tier (→ 29 commands). Router now lists `/linkedin:firsthour`; `calendar` cross-links to it. **Minor / additive** — no command removed, renamed, or behavior-changed; reload registers the two new commands. | | **4.0.0** | 2026-05-30 | **Audit-remediation release (Voyage Phase 0–3).** A critical self-review found overclaiming (tracking/analytics/review-independence the plugin couldn't deliver), dormant capability (11 agents never invoked by any command), and structural rot (a dead lint, a self-contradicting algorithm claim, an unpublishable model brand/date in user copy). The fix: every claim made honest or removed; **all 11 orphaned agents wired** (no deletions → 19 agents); new **`/linkedin:firsthour`** command (→ 27 commands) + short-form de-AI gate + video quality gate; `post-feedback-monitor` → Opus; honest newsletter-distribution / profile-SEO / outreach surfaces; **algorithm signals reconciled to one sourced statement** (no model name/date) with `references/algorithm-signals-reference.md` as the single source of truth; analytics fresh-clone crash fixed; **voice-profile leak closed** (placeholder + sentinel + gitignore); structure lint rebuilt with version/count/stat/model-consistency + render-chain-propagation + `$`-safety guards (post-release remediation S8–S13); 7-gate long-form review stack measured → no gate trimmed. **Breaking — reinstall/reload required** for the newly-wired agents; consolidates the v3.0.0 identity break. | | **3.1.0** | 2026-05-29 | Adversarial review package (Endring 9). New **`/linkedin:headless-review`** command + **Step 6.5 (headless-review)** in `/linkedin:newsletter`: three cold/headless archetypes — **`content-reviewer`** (argument), **`language-reviewer`** (Norwegian), **`fact-reviewer`** (cold re-verification + pivot-risk) — plus `persona-reviewer` (resonance/conversion), all with NO drafting-session context. New **`/linkedin:pivot`** command + pivot-detection gate (> 20 % word-count / > 2 new sections re-opens cleared gates before lock). **Per-artifact personas** (`articles.NN.personas`). Pipeline 15 → 16 phases; 24 → 26 commands; 16 → 19 agents; additive `personas`/`pivots`/`headlessReview` state. Backward-compatible; reload required for the new agents. | | **3.0.0** | 2026-05-29 | **Renamed** `linkedin-thought-leadership` → `linkedin-studio` ("LinkedIn Thought Leadership" → **LinkedIn Studio**). Breaking (slug + agent namespace `linkedin-studio:` + runtime state path `~/.claude/linkedin-studio.local.md` all change; reinstall required, state migrated in place), but **functionality byte-identical to v2.4.0**. The `/linkedin:*` commands are unchanged (frontmatter-namespaced, slug-independent). Catch-all skill dir renamed to match (`skills/linkedin-studio/`); the five functional skills unchanged. | diff --git a/plugins/linkedin-studio/commands/calendar.md b/plugins/linkedin-studio/commands/calendar.md index ed9ea23..0baa665 100644 --- a/plugins/linkedin-studio/commands/calendar.md +++ b/plugins/linkedin-studio/commands/calendar.md @@ -132,7 +132,11 @@ writeState(content => updatePostTracking(content, { ``` Replace placeholders with actual post data from the published post. -**3e. First-hour battle plan.** Show after marking: +**3e. First-hour battle plan.** Show the lightweight in-flow nudge below after marking. +For the **full worked sprint** — timestamped engagement targets (whales / inner-circle / +ICPs), draft self-comments + CEA replies in voice, and a minute-by-minute timeline — run +`/linkedin:firsthour`, which persists the plan to state and hands off to +`post-feedback-monitor`. The checklist here is the quick version: ``` Post marked as published! Here's your first-hour plan: diff --git a/plugins/linkedin-studio/commands/create.md b/plugins/linkedin-studio/commands/create.md new file mode 100644 index 0000000..99549da --- /dev/null +++ b/plugins/linkedin-studio/commands/create.md @@ -0,0 +1,66 @@ +--- +name: linkedin:create +description: | + Content-creation front-door — one guided entry for when you want to make something + but haven't picked the format yet. Asks what you want to create and routes you to the + command that owns that format (post, quick, react, carousel, video, multiplatform, + batch, or the long-form newsletter). It does NOT draft anything itself — it hands off + to the command that owns the work, so each format keeps its own voice rules and + quality gates. + Triggers on: "create", "make something", "create content", "what should I make", + "new content", "help me create", "i want to post something", "linkedin create". +allowed-tools: + - Read + - Glob + - AskUserQuestion +--- + +# Create — Content Front-Door + +You are the entry point for the **Create** journey. The user wants to make content but +may not know which format or command fits. Your job: identify the intent in one +question and route to the command that owns the work. **You do not draft here.** + +## Step 0: Quick context (optional) + +If `~/.claude/linkedin-studio.local.md` exists, you may glance at the planned next +topic and recent pillars to make a smarter suggestion — but keep it to one line and +do not block on it. + +## Step 1: Identify what they want to create + +If the user's message already names a format (e.g. "a carousel about X", "react to +this URL", "a video script") OR hands you a URL, skip the question and route directly +per the map below. + +Otherwise use `AskUserQuestion` — **"What do you want to create?"** + +1. **A post** — a full, substantial post (angle → draft → refine) → `/linkedin:post` +2. **A quick post** — a fast 5-minute post or a templated post-type → `/linkedin:quick` +3. **React to something** — turn a URL / article / news into a post → `/linkedin:react` +4. **A carousel** — a multi-slide / document post → `/linkedin:carousel` +5. **A video script** — talking-head / screen-recording / slideshow → `/linkedin:video` +6. **Adapt existing content** — for Twitter/X, slides, YouTube → `/linkedin:multiplatform` +7. **A whole week** — batch 3–5 posts in one session → `/linkedin:batch` +8. **Long-form** — newsletter edition, essay, or series article → `/linkedin:newsletter` + +## Step 2: Route + +State the chosen command and one line of why, then **proceed into that command's +workflow** (route to `/linkedin:` and begin it). Do NOT inline or duplicate +the target's steps — each creation command owns its own workflow, voice rules, and +quality gates. + +| Intent | Command | +|--------|---------| +| Substantial post | `/linkedin:post` | +| Quick / templated post | `/linkedin:quick` | +| React to a URL / article | `/linkedin:react` | +| Carousel / document | `/linkedin:carousel` | +| Video script | `/linkedin:video` | +| Cross-platform adaptation | `/linkedin:multiplatform` | +| A full week | `/linkedin:batch` | +| Long-form (newsletter / essay / series) | `/linkedin:newsletter` | + +**Long-form lock:** newsletters, essays, and series articles are owned end-to-end by +`/linkedin:newsletter` — the single long-form entry point. Never draft long-form here. diff --git a/plugins/linkedin-studio/commands/linkedin.md b/plugins/linkedin-studio/commands/linkedin.md index 67da103..8e07379 100644 --- a/plugins/linkedin-studio/commands/linkedin.md +++ b/plugins/linkedin-studio/commands/linkedin.md @@ -58,99 +58,92 @@ If queue is empty: "No posts scheduled. Run /linkedin:batch to plan your week." ## Available Commands -Present these options to the user: +LinkedIn Studio is organized as **five journeys**. Each journey has a **front-door** +command (start here if you're unsure) plus the focused commands it routes to. Type a +front-door when you know the journey but not the exact command; type a specific command +directly when you do. -### Getting Started +### 🟢 Start — get set up and publish your first post | Command | Purpose | |---------|---------| -| `/linkedin:onboarding` | Full onboarding wizard — profile, setup, and first post in one flow | +| `/linkedin:onboarding` | **Front-door.** Full guided flow: profile → setup → first post | +| `/linkedin:setup` | Populate voice samples, case studies, frameworks, and audience data | | `/linkedin:first-post` | First-post accelerator — zero to published in under 10 minutes | -| `/linkedin:setup` | Guided setup to populate empty asset templates with your real voice, case studies, and audience data | -### Content Creation +### ✍️ Create — make content | Command | Purpose | |---------|---------| +| `/linkedin:create` | **Front-door.** Tells you which creation command fits, then routes you | | `/linkedin:post` | Full post creation with angle selection, format choice, and refinement | +| `/linkedin:quick` | Fast 5-minute post (3-line formula) + the 8 post-type templates | | `/linkedin:react` | React to a URL (article, news, research) and turn it into a post | -| `/linkedin:quick` | Fast 5-minute post using the 3-line formula (also the 8 post-type templates) | -| `/linkedin:pipeline` | Full end-to-end workflow from idea to post-publish analysis | -| `/linkedin:carousel` | Create structured multi-slide carousel with visual layout guidance | -| `/linkedin:video` | Create video scripts with hook, body, CTA, captions, and thumbnail suggestions | -| `/linkedin:newsletter` | Long-form orchestrator — newsletter editions, essays, series articles (research → draft → fact-check → persona-review → lock → delivery). The single long-form entry point | -| `/linkedin:headless-review` | Cold adversarial review of a FROZEN long-form draft (argument, language, facts, reader-fit) — run before lock, ideally in a fresh session for max independence | -| `/linkedin:pivot` | Re-open a long-form edition after a substantive late change so cleared gates (fact-check → editorial → persona → headless) re-run before lock | +| `/linkedin:carousel` | Structured multi-slide carousel with visual layout guidance | +| `/linkedin:video` | Video scripts with hook, body, CTA, captions, and thumbnail suggestions | +| `/linkedin:multiplatform` | Adapt content for Twitter/X, slides, YouTube (long-form → newsletter) | | `/linkedin:batch` | Create a full week of content in one session | -| `/linkedin:calendar` | View and manage your post scheduling queue + run the publish action (mark a scheduled post as published) | +| `/linkedin:pipeline` | End-to-end single-post workflow (idea → draft → schedule → analyze) | +| `/linkedin:newsletter` | **Long-form spine.** Newsletter editions, essays, series articles. The single long-form entry point | +| `/linkedin:headless-review` | Cold adversarial re-read of a FROZEN long-form draft before lock (ideally in a fresh session) | +| `/linkedin:pivot` | Re-open a long-form edition after a late change so cleared gates re-run | -### Strategy & Optimization +### 📅 Engage — ship and work the first hour | Command | Purpose | |---------|---------| -| `/linkedin:strategy` | Growth strategy + authority building (phase-specific guidance, trajectory adjustments, signature content for Phase 2+) | -| `/linkedin:profile` | profile/topic-relevance optimization checklist | -| `/linkedin:analyze` | Analyze content performance or troubleshoot issues | -| `/linkedin:ab-test` | Design and manage A/B tests for content optimization | -| `/linkedin:audit` | Quarterly content strategy audit | -| `/linkedin:competitive` | Competitive analysis of other thought leaders | +| `/linkedin:calendar` | View/manage the queue + the publish action (mark a scheduled post as published) | +| `/linkedin:firsthour` | Post-publish first-hour sprint: timestamped targets + draft comments + timeline | + +After publishing, the `post-feedback-monitor` agent tracks performance in the critical +first 48 hours, detects anomalies, and advises real-time interventions. + +### 📊 Measure — understand performance + +| Command | Purpose | +|---------|---------| +| `/linkedin:measure` | **Front-door.** Routes you to the right analytics command | | `/linkedin:import` | Import LinkedIn CSV exports for analytics | -| `/linkedin:report` | Generate weekly performance report with trends and alerts | +| `/linkedin:report` | Weekly/monthly performance report with trends and alerts | +| `/linkedin:analyze` | Troubleshoot performance (reach dropped, low engagement) | +| `/linkedin:audit` | Quarterly content strategy audit | +| `/linkedin:ab-test` | Design and manage A/B tests for content optimization | -### Post-Publish Monitoring - -| Agent | Purpose | -|-------|---------| -| `post-feedback-monitor` | Monitor post performance in the critical first 48 hours, detect anomalies, and get real-time intervention advice | - -### Growth & Monetization +### 🚀 Grow — authority, reach, and revenue | Command | Purpose | Unlocks at | |---------|---------|-----------| +| `/linkedin:strategy` | **Front-door.** Phase roadmap, trajectory, authority + signature content | Any phase | +| `/linkedin:profile` | profile/topic-relevance optimization checklist | Any phase | +| `/linkedin:competitive` | Competitive analysis of other thought leaders | Any phase | | `/linkedin:monetize` | Monetization strategy (lead magnets, consulting funnel, pricing) | ~1K followers | -| `/linkedin:outreach` | Outreach orchestrator — collaborations and speaking opportunities (CFPs, partner pitches) | ~1K followers | -| `/linkedin:multiplatform` | Adapt content for Twitter/X, slides, YouTube (short-form/cross-format; long-form → `/linkedin:newsletter`) | Any phase | +| `/linkedin:outreach` | Collaborations and speaking opportunities (CFPs, partner pitches) | ~1K followers | -**Gating rule:** the "Unlocks at ~1K followers" commands are deliberately -listed but soft-gated — they work at any follower count, but their value -compounds once a profile has the audience scale and authority signal that -makes lead magnets, partnerships, and speaking pitches realistic. Below -~1K followers the router will note the threshold and suggest -`/linkedin:strategy` first. +**Gating rule:** the "Unlocks at ~1K followers" commands are deliberately listed but +soft-gated — they work at any follower count, but their value compounds once a profile +has the audience scale and authority signal that makes lead magnets, partnerships, and +speaking pitches realistic. Below ~1K followers the router notes the threshold and +suggests `/linkedin:strategy` first. (`/linkedin:competitive` is **not** gated — +competitive analysis helps at any stage, especially early positioning.) ## Ask the User -Use AskUserQuestion to ask: +Show the five journeys as a menu and let the user pick a journey or jump straight to a +command. Lead with the front-doors. **What would you like to do?** -0. **Onboarding wizard** — Just installed? Full guided flow: profile → setup → first post -1. **Setup & personalize** — Guided setup to populate voice, case studies, frameworks, and audience data -2. **Create a post** — Full post workflow with angle selection -3. **React to a URL** — Turn an article/news into a post -4. **Quick post** — Post something fast (5 min) -5. **Full pipeline** — End-to-end: idea → draft → optimize → publish -6. **Batch create** — Create a full week of content in one session -7. **Use a template** — Browse proven post templates -8. **View calendar** — See scheduled posts and manage queue -9. **Plan content** — Weekly/monthly content planning -10. **Growth strategy & authority** — Plan growth, build signature content, compound authority -11. **Optimize profile** — topic-relevance profile audit -12. **Audit content** — Review what's working and what's not -13. **Competitive analysis** — Learn from other thought leaders -14. **Monetize** — Revenue strategies for thought leadership -15. **Outreach** — Collaborations and speaking opportunities (CFPs, partner pitches, joint content) -16. **Multi-platform** — Adapt content for other platforms -17. **Troubleshoot** — My content isn't performing well -18. **Monitor post** — Check how my latest post is performing (first 48 hours) -19. **A/B test** — Design, track, or analyze content experiments -20. **Create a video** — Script for talking head, screen recording, or slideshow -21. **Mark published** — I just published a scheduled post -22. **Write long-form** — Newsletter edition, essay, or series article (`/linkedin:newsletter`) -23. **Cold review (headless)** — Independent adversarial re-read of a frozen long-form draft before lock (`/linkedin:headless-review`) -24. **Pivot a long-form edition** — Re-open cleared gates after a late substantive change (`/linkedin:pivot`) +- **🟢 Start** — `onboarding` (full wizard) · `setup` · `first-post` +- **✍️ Create** — `create` (guided) · post · quick · react · carousel · video · multiplatform · batch · pipeline · newsletter · headless-review · pivot +- **📅 Engage** — `calendar` (queue + mark published) · `firsthour` (first-hour sprint) +- **📊 Measure** — `measure` (guided) · import · report · analyze · audit · ab-test +- **🚀 Grow** — `strategy` (roadmap) · profile · competitive · monetize ⚿ · outreach ⚿ -Based on their answer, guide them to the appropriate command or invoke it directly. +If they name a command, route to it. If they name a journey but not a command, route to +that journey's front-door (`create` / `measure` / `onboarding` / `strategy`); for Engage, +ask whether they're scheduling (`calendar`) or have just published (`firsthour`). Use +`AskUserQuestion` for any sub-choice (≤4 options per question). Based on their answer, +guide them to the appropriate command or invoke it directly. ## If They Have Specific Content @@ -198,6 +191,9 @@ If the user's intent is clear from context: - Mentions "status" or "on track" → Route to `/linkedin:calendar` (plan-vs-queue diff) - Mentions "repurpose" or "reuse" → Suggest `content-repurposer` agent - Mentions "video" or "video script" or "film" or "record" or "talking head" or "screen recording" or "slideshow video" → Route to `/linkedin:video` +- Mentions "create" or "make something" or "what should I make" or "new content" (no specific format named) → Route to `/linkedin:create` (the Create front-door) +- Mentions "measure" or "how am I doing" or "my performance" or "performance overview" or "how are my posts doing" → Route to `/linkedin:measure` (the Measure front-door) +- Mentions "first hour" or "I just posted" or "reply loop" or "work my post" or "engage now" → Route to `/linkedin:firsthour` (the worked first-hour sprint plan — distinct from `/linkedin:calendar`'s mark-as-published action) - Has a URL to react to → Route to `/linkedin:react` - Has substantial content to convert → Route to `/linkedin:post` diff --git a/plugins/linkedin-studio/commands/measure.md b/plugins/linkedin-studio/commands/measure.md new file mode 100644 index 0000000..47b9d32 --- /dev/null +++ b/plugins/linkedin-studio/commands/measure.md @@ -0,0 +1,51 @@ +--- +name: linkedin:measure +description: | + Performance front-door — one guided entry for when you want to understand how you're + doing on LinkedIn. Asks what you need and routes you to the command that owns that + analysis (import, report, analyze/troubleshoot, audit, or A/B test). It does NOT + crunch numbers itself — it hands off to the command that owns the work. + Triggers on: "measure", "how am I doing", "my performance", "show my analytics", + "performance overview", "how are my posts doing", "linkedin measure". +allowed-tools: + - Read + - Glob + - AskUserQuestion +--- + +# Measure — Performance Front-Door + +You are the entry point for the **Measure** journey (the measure → improve loop). The +user wants insight into performance but may not know which command fits. Identify the +intent in one question and route. **You do not run the analysis here.** + +## Step 0: Quick context (optional) + +If `assets/analytics/` holds imported data, you may note "last import: [date]" in one +line so the user knows whether a fresh import is needed first. Do not block on it. + +## Step 1: Identify what they need + +Use `AskUserQuestion` — **"What do you want to do?"** + +1. **Import new data** — load a LinkedIn analytics CSV export → `/linkedin:import` +2. **See a report** — weekly/monthly numbers, trends, top performers → `/linkedin:report` +3. **Diagnose a problem** — reach dropped / low engagement, what's wrong → `/linkedin:analyze` +4. **Strategy audit** — periodic top/bottom posts, topic & format mix review → `/linkedin:audit` +5. **A/B test** — design, log, or review a content experiment → `/linkedin:ab-test` + +## Step 2: Route + +State the chosen command and one line of why, then **proceed into that command's +workflow**. Do NOT duplicate the analysis logic — each command owns it. + +| Intent | Command | +|--------|---------| +| Import a CSV export | `/linkedin:import` | +| Weekly / monthly report | `/linkedin:report` | +| Troubleshoot performance | `/linkedin:analyze` | +| Quarterly strategy audit | `/linkedin:audit` | +| Content experiment | `/linkedin:ab-test` | + +**Order note:** reporting needs imported data. If the user wants a report but nothing +has been imported yet, route to `/linkedin:import` first, then `/linkedin:report`. diff --git a/plugins/linkedin-studio/commands/onboarding.md b/plugins/linkedin-studio/commands/onboarding.md index f6f5151..3116329 100644 --- a/plugins/linkedin-studio/commands/onboarding.md +++ b/plugins/linkedin-studio/commands/onboarding.md @@ -3,7 +3,7 @@ name: linkedin:onboarding description: | Multi-step onboarding wizard that guides new users through profile → setup → first-post as one cohesive flow. Designed for users who have just installed the plugin and want a - single guided path instead of navigating 27 commands on their own. + single guided path instead of navigating 29 commands on their own. Triggers on: "onboarding", "get started", "new user", "setup wizard", "start from scratch", "just installed", "how do I start", "walk me through", "linkedin onboarding". allowed-tools: @@ -214,4 +214,4 @@ First post: [Published DATE / Pending — run /linkedin:first-post] - `/linkedin:batch` — Plan a full week of content in one session - `/linkedin:react` — Turn articles and news into posts - `/linkedin:strategy` — Growth strategy tailored to your follower level -- `/linkedin` — See all 27 commands anytime +- `/linkedin` — See all 29 commands anytime diff --git a/plugins/linkedin-studio/docs/remediation/command-rationalization.md b/plugins/linkedin-studio/docs/remediation/command-rationalization.md new file mode 100644 index 0000000..1733ffe --- /dev/null +++ b/plugins/linkedin-studio/docs/remediation/command-rationalization.md @@ -0,0 +1,375 @@ +# Command-Surface Rationalization — Step 14a + +_Remediation Voyage, command-surface pass. Independent, COLD read of all 27 +command files in `commands/*.md` (full bodies, not just frontmatter), with the +router (`commands/linkedin.md`), `CLAUDE.md`, and `README.md` as the map of the +intended surface. Written 2026-05-30. ANALYSIS ONLY — no command file was edited, +merged, deleted, or renamed; the single deliverable is this document._ + +> **Correction (post-review, 2026-05-30).** The first pass omitted `multiplatform` +> (covered 26/27) and mis-stated two group-header counts; it also raised a +> `competitive` 1K-gating inconsistency that did **not** survive verification. All +> three are fixed below: `multiplatform` now has its own entry (Group B), the Group +> C/F headers read 3/5, and the `competitive` nit is withdrawn (CLAUDE.md `:64`, +> README `:222`, the router, and the command body all leave it ungated). Net finding +> is unchanged — **keep 27, 0 merge, 0 cut** — now genuinely covering all 27. + +## Method note + +The fixtures here are **the command files themselves** — each command's purpose, +its full workflow body, its frontmatter `description` (the trigger-phrase set that +governs how it gets invoked), and the sibling surfaces it touches. The discipline +is borrowed verbatim from `docs/remediation/overlap-measurement.md` (whose SUBJECT +is the 7 long-form review *agents*, not the commands — kept strictly separate +here): for every overlap I apply the **subsumption test** — _does either command's +catch-set / surface fully contain the other's?_ — and the trim rule: **merge or +cut ONLY where a command catches nothing a sibling doesn't; if the redundancy is +justified, record it and KEEP; if I can't decide, record "inconclusive — retain +pending operator view" and do NOT trim.** Default-to-keep under uncertainty. This +plugin already over-grew once (24→27), so the bias is honest-surface, not +consolidation-for-its-own-sake and not growth-for-its-own-sake. Every +recommendation is grounded in a citable trigger set, workflow step, or sibling +surface — no taste-based cuts. + +A leverage note used throughout: a command's **invocation leverage** is the +product of (a) trigger-phrase breadth in its `description` and (b) how likely a +user is to actually type the intent when a *broader* command already catches it. +A command whose entire trigger set is a strict subset of a broader command's, and +which adds no unique workflow, is the classic merge/cut candidate. + +--- + +## Group A — Entry / router (1) + +### `linkedin` (router) + +- **Purpose** — Status line (weekly progress, streak, follower phase) + upcoming/overdue queue + a menu/router that disambiguates to the right subcommand. +- **Overlap with siblings** — Structurally overlaps *every* command (it lists them), but that is its job, not redundancy. No subsumption either way: the router holds no content-production surface, the leaf commands hold no menu/status surface. +- **Invocation leverage** — Highest-traffic entry point: triggers on the bare `"linkedin"`, `"/linkedin"`, `"linkedin help"`. This is the front door. +- **Recommendation** — **keep.** +- **Rationale** — Structural and the single highest-leverage surface. Judged on grouping honesty: the grouping is broadly honest (Getting Started / Content Creation / Strategy / Post-Publish / Growth+Monetization), and it correctly soft-gates the 1K commands. One honest-surface nit for the operator (not a cut): it lists `post-feedback-monitor` as an *agent* under "Post-Publish Monitoring" but does not list the new `/linkedin:firsthour` command in that section — the router's own map is one release behind the surface it routes to. + +--- + +## Group B — Short-form creation (7) + +### `post` + +- **Purpose** — Full interactive post creation (angle → format → draft → quality/de-AI gate → refine), targeting 1,200–1,800-char substantial posts. +- **Overlap with siblings** — `quick` (short 150–500-char path), `react` (URL-sourced), `pipeline` (post + schedule + monitor). Subsumption test: `post` does NOT subsume `quick` (different length band + 3-line formula + 8 templates) and `quick` does not subsume `post` (no refinement cycle, no `content-optimizer` delegation). `post` is the body that `pipeline` wraps but `pipeline` adds scheduling/queue/first-hour that `post` lacks; neither subsumes the other. +- **Invocation leverage** — Very high; broad trigger set ("create linkedin post", "write a post", "turn this into a post"). Core entry point alongside `quick`. +- **Recommendation** — **keep.** +- **Rationale** — Distinct length/depth tier with a unique refinement+optimizer surface; one of the two highest-traffic creation commands. + +### `quick` + +- **Purpose** — 5-minute 3-line-formula post (150–500 chars) AND the home of the 8 post-type templates (reaction/tip/observation/hot-take/failure/question/curation/one-liner). +- **Overlap with siblings** — `post` (longer tier), `first-post` (which explicitly reuses `quick`'s 3-line formula). Subsumption test: `quick` uniquely owns the template library and the short-length band; `post` owns the long band. The two are explicitly cross-referenced ("When to Upgrade → `/linkedin:post`"). No subsumption. +- **Invocation leverage** — High; broad trigger set including the absorbed `templates` intent ("post template", "give me a template", "fill in the blank post"). This absorption was the v2.0.0 consolidation — confirmed intact in the body (Step 1 hosts the template library). +- **Recommendation** — **keep.** +- **Rationale** — Holds the unique short-form length tier + the entire template surface that `templates` was folded into. Cutting it would orphan the templates. + +### `react` + +- **Purpose** — URL-to-post pipeline: fetch external content (article/news/research/YouTube), extract, pick a reaction angle, draft in voice; includes a multi-URL comparison path. +- **Overlap with siblings** — `post` (could in principle ingest a URL via WebFetch — `post` Step 1 even mentions "If they provide a URL, use WebFetch"). Subsumption test: `react` has a unique surface `post` lacks — the multi-source comparison/synthesis path (Steps 1b–8b), the content-type→angle table tuned to *reactions*, and the "react, don't summarize / your take is the hook" discipline. `post` does not subsume that. `react` does not subsume `post` (no general-topic drafting). +- **Invocation leverage** — High and unambiguous: "react to this", "this article", "this url", "share this news". The router routes any bare URL here. A user with a link will not naturally type `/linkedin:post`. +- **Recommendation** — **keep.** +- **Rationale** — Genuinely unique URL-ingestion + multi-source-synthesis surface and a distinct, frequently-typed intent. + +### `pipeline` + +- **Purpose** — End-to-end lifecycle orchestrator: ideation → draft → optimize → **schedule/queue** → 5x5x5 pre-engagement → publish → first-hour → 48h analysis. +- **Overlap with siblings** — This is the densest overlap node. It re-does `post`'s drafting (Step 2), `calendar`'s queueing (Step 4 Option 3 calls the same `queue-manager.mjs` `queueAdd`), the 5x5x5/first-hour plan (now also `firsthour`'s job), and points to `analyze` at the end. Subsumption test: no *single* sibling subsumes `pipeline` — it is the only command that strings the whole chain into one guided run with explicit step gates. But `pipeline` is itself **a thinner re-statement of `post` + `calendar` (+ `firsthour`)**: every individual capability it offers exists, more developed, in those leaf commands (its draft step is a condensed `post`; its schedule step is a condensed `calendar`; its first-hour step is a condensed `firsthour`). +- **Invocation leverage** — Moderate-to-low as a *natural* type. Trigger set ("pipeline", "full workflow", "end to end", "idea to post") is narrow and jargon-y; in practice a user wanting to create+schedule will type `/linkedin:post` then `/linkedin:calendar`. It reads as a workflow-orchestrator showcase more than a daily entry point. +- **Recommendation** — **keep** (with a flag) — leaning **inconclusive**. +- **Rationale** — No clean subsumption (it uniquely *sequences* the chain), so the trim rule says retain. But it is the surface most worth the operator's scrutiny: its value is the *orchestration*, and that value erodes as `post`/`calendar`/`firsthour` each grow richer than `pipeline`'s inlined condensations. Flagged in the honest-uncertainty list. If the operator ever wants to shrink the surface, this is the first place a defensible merge could be argued (→ `post`, preserving the explicit schedule+first-hour hand-off chain) — but the evidence does not *compel* it. + +### `carousel` + +- **Purpose** — Structured slide-by-slide carousel/document generator with template selection, per-slide copy, caption, de-AI gate, and optional mcp-image slide generation. +- **Overlap with siblings** — `multiplatform` (has a "LinkedIn → Presentation Slides" template) and `post` (notes "could also work as a carousel — run `/linkedin:carousel`"). Subsumption test: `multiplatform`'s slide template is a 10-line generic outline; `carousel` is a full format engine (5 templates, per-slide char rules, mcp-image generation, carousel quality checklist). `multiplatform` does not remotely subsume it. `carousel` owns the highest-engagement organic format end-to-end. +- **Invocation leverage** — High and specific: "carousel", "slide deck", "pdf post", "swipe post", "document post". A distinct format intent users type directly. +- **Recommendation** — **keep.** +- **Rationale** — Deep, unique format surface with image generation; not subsumed by anything. + +### `video` + +- **Purpose** — Video script generator (talking-head/screen-rec/slideshow, 30s–2min) with pacing/visual/energy cues, captions, thumbnail, first comment, a video quality gate, and delegation to `video-scripter`. +- **Overlap with siblings** — `multiplatform` ("LinkedIn → YouTube Script" template). Subsumption test: `multiplatform`'s YouTube template is a generic timing outline; `video` is a full LinkedIn-native scripting engine (word-budget math, muted-autoplay test, completeness gate, `video-scripter` agent). No subsumption — and they target different platforms (LinkedIn-native vs. YouTube adaptation). +- **Invocation leverage** — High and specific: "video script", "talking head script", "record a video". Direct format intent. +- **Recommendation** — **keep.** +- **Rationale** — Unique LinkedIn-native video surface + dedicated agent; the `multiplatform` overlap is a thin adaptation template, not a substitute. + +### `multiplatform` + +- **Purpose** — Adapt existing LinkedIn content for other platforms: a Twitter/X thread, a generic presentation deck, or a YouTube script. Long-form is explicitly routed out to `/linkedin:newsletter`. +- **Overlap with siblings** — `carousel` (its "Presentation Slides" template) and `video` (its "YouTube Script" template). Subsumption test: `multiplatform`'s slide/YouTube blocks are ~10-line generic outlines, while `carousel` is a full LinkedIn-native carousel engine and `video` a full LinkedIn-native scripting engine — neither subsumes the other, and they target different outputs (LinkedIn-native vs. Twitter / YouTube / generic deck). Critically, the **Twitter/X-thread** adaptation path exists in **no other command**. No subsumption. +- **Invocation leverage** — Moderate; "adapt for twitter", "cross-post", "repurpose for", "turn into thread". A distinct cross-platform intent, but lower-traffic than the native creation commands. +- **Recommendation** — **keep** (flag: **develop-candidate**). +- **Rationale** — Unique cross-platform / Twitter surface; subsumption fails as a cut. But it is the **thinnest** command in the set — the only one that inlines static templates instead of delegating, where the `content-repurposer` agent already exists to power a richer adaptation. Keep now; a future *develop* (wire `content-repurposer`, deepen the per-platform output) is the defensible improvement, not a merge/cut. + +--- + +## Group C — Long-form (3) + +### `newsletter` + +- **Purpose** — The single long-form orchestrator: a fixed 16-phase, multi-session pipeline (research → skeleton/spine gates → draft → fact-check → editorial → persona sweep → headless review → visual assets → lock → hook gate → schedule) with maintained edition-state. +- **Overlap with siblings** — `headless-review` and `pivot` are *phases/companions* of this command (Step 6.5 and the lock-precondition re-open). Subsumption test: `newsletter` invokes the headless package inline (Step 6.5) and runs the pivot heuristic as a lock precondition (Step 8) — so it functionally *contains* both. But containment ≠ the companions are redundant (see their entries: the value is fresh-session isolation and a named re-open ritual, which `newsletter` cannot itself provide). No short-form command overlaps it; `multiplatform` explicitly routes all long-form here. +- **Invocation leverage** — High for its niche; the sole entry for "newsletter", "long-form", "essay", "series article". +- **Recommendation** — **keep.** +- **Rationale** — The entire long-form spine; nothing else does this and everything long-form routes to it. + +### `headless-review` + +- **Purpose** — Cold/adversarial review package run on a FROZEN draft with starved context — `content-reviewer` + `language-reviewer` + `fact-reviewer` + `persona-reviewer` (resonance/conversion), consolidated into one operator-gated report. Standalone surface for `newsletter` Step 6.5. +- **Overlap with siblings** — `newsletter` (Step 6.5 fans the same package inline). Subsumption test: this is the calibration case (c) the brief flags. Does `newsletter` Step 6.5 subsume `headless-review`? **No** — and the command's own body states *why*: the cardinal value is **Layer 1 fresh-session isolation** ("the parent itself then has no drafting transcript"). When the package runs inline inside the drafting `newsletter` session it carries exactly the framing-bias the package exists to eliminate. The standalone command is the *only* way to get a genuinely cold parent context. So the standalone surface has a unique catch (true independence) the inline phase structurally cannot. +- **Invocation leverage** — Moderate; specific trigger set ("headless review", "cold review", "adversarial review", "review the frozen draft"). Niche but real, and the recommended path is to type it *in a fresh session* — which is precisely an action a phase cannot perform. +- **Recommendation** — **keep** — justified standalone, not merely-as-a-phase. +- **Rationale** — Subsumption fails in the independence dimension: the fresh-session parent is a capability the `newsletter` phase cannot replicate. This is the "redundancy is justified — record and keep" case, mirroring the agent-overlap finding for the cold trio in `overlap-measurement.md`. + +### `pivot` + +- **Purpose** — A named ritual to re-open an already-cleared long-form edition after a substantive late change: logs `pivots[]`, resets `currentPhase`, un-locks, invalidates downstream verdicts, marks which gates must re-pass. Includes the >20%/>2-section heuristic. +- **Overlap with siblings** — `newsletter` (which runs the same heuristic as a Step 8 lock precondition and owns the gate re-runs). Subsumption test (calibration case c): does `newsletter` subsume `pivot`? **Partially but not fully.** `newsletter` Step 8 *detects* drift and *stops* the lock, but it explicitly **points the operator to `/linkedin:pivot`** to perform the state surgery (log the pivot entry, reset phase, un-lock, invalidate verdicts). `pivot` does NOT run the gates ("Do not run the gates yourself — `/linkedin:newsletter` owns the pipeline"); `newsletter` does NOT perform the re-open bookkeeping. The two are complementary halves of one ritual — clean separation, no subsumption. +- **Invocation leverage** — Low-to-moderate; narrow trigger set ("pivot", "re-open edition", "added a section", "changed the angle"). Rarely hit, but when hit it does deterministic state work no other command does. +- **Recommendation** — **keep.** +- **Rationale** — Holds a unique state-mutation surface (the re-open bookkeeping) that `newsletter` deliberately delegates out; not subsumed. Lower-traffic, but the trim rule retains a non-redundant specialist. + +--- + +## Group D — Onboarding / setup (3) + +### `onboarding` + +- **Purpose** — Multi-step wizard chaining profile → personalization → first-post as one guided flow for brand-new installs, with an "already onboarded" short-circuit. +- **Overlap with siblings** — `setup` (Phase 2 is a condensed `setup`) and `first-post`/`profile` (Phases 1 and 3 route to them). Subsumption test (calibration case f): does `setup` + `first-post` subsume `onboarding`? **No.** `onboarding` uniquely provides the *cohesive single path* ("a single guided path instead of navigating 27 commands") and the cross-phase state logic (already-onboarded detection, score-gated branching, "what's next — your first week"). Notably it does NOT invoke the sub-commands directly — it tells the user to run them — so it is an orchestration/triage layer, not a duplicate of their bodies. +- **Invocation leverage** — Moderate; trigger set heavily overlaps `first-post` and `setup` ("get started", "just installed", "walk me through"). The router lists it first. Real risk: a "get started" user could land on `first-post` instead — but the intent ("walk me through *everything*") is distinct from "help me publish my first post". +- **Recommendation** — **keep.** +- **Rationale** — Unique cohesive-wizard surface + cross-phase triage state; it orchestrates rather than duplicates `setup`/`first-post`. No subsumption. + +### `setup` + +- **Purpose** — Guided personalization: computes the 8-category personalization score and runs 6 sub-workflows (voice samples → `voice-trainer`, case study, framework, post analysis, demographics, user profile) to populate asset templates. +- **Overlap with siblings** — `onboarding` Phase 2 (a 2-category condensation of this). Subsumption test: `onboarding` only ever touches the top-2 weighted categories and explicitly defers ("I'll run `/linkedin:setup` for the full setup"); `setup` owns all 8 categories, the full score dashboard, and the `voice-trainer` delegation. `onboarding` does not subsume it. +- **Invocation leverage** — High; broad trigger set ("setup", "personalize", "personalization score", "configure plugin", "my score", "fill in assets"). Also the router's destination for any score/asset-completeness query. +- **Recommendation** — **keep.** +- **Rationale** — The canonical, full-depth personalization surface; `onboarding`'s version is a deliberate condensation that defers back here. + +### `first-post` + +- **Purpose** — First-post accelerator: maximum hand-holding, voice quick-check, simple topic pick, 3-line draft (reuses `quick`'s formula), 4-item quality check, sets `first_post_date`, includes a guard that redirects returning users to `post`/`quick`. +- **Overlap with siblings** — `quick` (whose 3-line formula it reuses) and `onboarding` Phase 3 (which routes here). Subsumption test: does `quick` subsume `first-post`? **No.** `first-post` adds first-timer-specific scaffolding `quick` lacks: the welcome/expectation framing, the voice-from-scratch setup branch (samples vs. 5 questions), the "exists > perfect" philosophy, the new-creator 90-day-boost messaging, and the returning-user guard. Its whole value is the zero-to-one onboarding hand-holding, not the drafting mechanics. +- **Invocation leverage** — Moderate; "first post", "never posted", "new to linkedin". A genuinely distinct beginner intent. +- **Recommendation** — **keep.** +- **Rationale** — Unique first-timer scaffolding + the `first_post_date` side-effect; not subsumed by `quick`'s formula it borrows. + +--- + +## Group E — Scheduling / publishing / post-publish (3) + +### `batch` + +- **Purpose** — Create a full week (3–5 posts) in one session from a theme/pillar: trend-spotter angles, `content-planner` plan, scheduling, draft files, queue entries, and a generated `.ics`. +- **Overlap with siblings** — `pipeline` (single post + schedule), `calendar` (queue management). Subsumption test: `batch` uniquely produces a *multi-post balanced week* with format/pillar rotation and `.ics` export; `pipeline` is single-post; `calendar` only manages an existing queue. No subsumption. +- **Invocation leverage** — High and specific: "batch content", "week of posts", "sunday prep". Distinct planning intent. +- **Recommendation** — **keep.** +- **Rationale** — Unique multi-post-week surface + iCal generation; the queue-write overlap with `calendar`/`pipeline` is shared *plumbing* (`queue-manager.mjs`), not a shared *surface*. + +### `calendar` + +- **Purpose** — View/manage the 14-day queue (reschedule/cancel/view-draft) AND host the **publish action** (mark-as-published → update queue+state → first-hour plan). Absorbed the former `publish` command (v2.0.0). +- **Overlap with siblings** — `firsthour` (its publish flow ends with a first-hour plan; `firsthour` is a richer standalone version), `pipeline`/`batch` (which write to the queue it reads). Subsumption test (calibration case d): does `calendar`'s publish action subsume `firsthour`? **No** — `calendar`'s first-hour block is a static 6-line checklist + an optional `post-feedback-monitor` hand-off; `firsthour` builds a *worked, timestamped plan with named targets and draft comments* via `engagement-coach`. Conversely `firsthour` does not manage the queue. No subsumption. +- **Invocation leverage** — Very high; very broad trigger set spanning both calendar ("schedule", "queue", "upcoming") and publish ("mark as published", "just published", "post is live"). High-traffic. +- **Recommendation** — **keep.** +- **Rationale** — Canonical queue+publish surface that already absorbed `publish`; its first-hour checklist is a lightweight in-flow nudge, distinct from `firsthour`'s full sprint planner. + +### `firsthour` + +- **Purpose** — Post-publish first-hour / reply-loop sprint: delegates to `engagement-coach` for a timestamped target list (whales/inner-circle/ICP), draft self-comments + CEA replies in voice, a minute-by-minute timeline; persists via `recordFirstHourPlan`; hands off to `post-feedback-monitor`. +- **Overlap with siblings** — `calendar`'s publish-action first-hour block, and the `engagement-coach` / `post-feedback-monitor` agents it delegates to. Subsumption test (calibration case d): `firsthour` is the orphan-wiring command for `engagement-coach` (its raison d'être per CLAUDE.md/README). It uniquely produces a *worked* plan (named targets + draft comments + clipboard copy + state persistence) that neither `calendar` (static checklist) nor `post-feedback-monitor` (48h monitoring, different time window) provides. The three form a relay (firsthour → coach for the plan → feedback-monitor for the marathon), not a redundancy. +- **Invocation leverage** — Moderate; "first hour", "I just posted", "reply loop", "work my post". A real, frequently-recurring moment (every publish), but its trigger phrases partly collide with `calendar`'s "just published". Some users will reach the lighter `calendar` first-hour block instead of `firsthour`. +- **Recommendation** — **keep** (with a note). +- **Rationale** — Distinct worked-plan surface + it is the wiring point for an otherwise-orphaned agent (`engagement-coach`); cutting it would re-orphan that agent and undo a core v4.0.0 remediation goal. Note for the operator: the trigger overlap with `calendar`'s publish action means the two should cross-link (calendar's first-hour block could point to `/linkedin:firsthour` for the full plan) — a wiring nit, not a merge. + +--- + +## Group F — Analytics (5) + +### `analyze` + +- **Purpose** — Performance *troubleshooting/diagnosis*: 6-symptom intake → diagnostic patterns → penalty checklist → severity assessment → 14-day recovery protocol. Optionally grounds in `analytics-interpreter` (interpret mode). +- **Overlap with siblings** — `audit`, `report`, `import` (calibration case h). Subsumption test: `analyze` is the only *diagnostic/recovery* surface (why is reach down, how do I recover). `report` is *reporting* (weekly numbers from imported data); `audit` is *strategy review* (quarterly top/bottom posts, topic mix); `import` is *ingestion*. Four different jobs on the analytics axis — none subsumes another; `analyze` even works with no imported data (self-report intake). +- **Invocation leverage** — High and distinct: "why isn't my content performing", "low reach", "reach dropped". A panic-moment intent users type directly. +- **Recommendation** — **keep.** +- **Rationale** — Unique troubleshooting/recovery surface; orthogonal to the other three analytics commands. + +### `audit` + +- **Purpose** — Periodic (quarterly) content-strategy audit: top/bottom performers, topic distribution vs. pillars, format mix, engagement trend, milestone progress, profile alignment — then routes the *fix* to `strategy` and the *profile deep-audit* to `profile`. +- **Overlap with siblings** — `analyze` (diagnosis), `report` (metrics), `strategy`/`profile` (which it explicitly defers to). Subsumption test: `audit` is the only *holistic strategy-review* surface (90-day lookback across topic/format/trajectory). It deliberately does NOT own trajectory prescription (delegates to `strategy`) or profile checklist (delegates to `profile`) — "audit names the gap; strategy prescribes the fix." That self-limiting design means it does not duplicate them; and `report`/`analyze` don't do the strategy lookback. No subsumption. +- **Invocation leverage** — Moderate; "content audit", "quarterly review", "review my content strategy". A periodic, distinct intent. +- **Recommendation** — **keep.** +- **Rationale** — Unique holistic-review surface with clean delegation boundaries to `strategy`/`profile`; not a duplicate of `analyze`/`report`. + +### `import` + +- **Purpose** — Ingest a LinkedIn analytics CSV (auto-detect from ~/Downloads, quick-import helper, the analytics-CLI npm-install fix), parse → JSON → anomaly detection → baseline update, then **delegates the analysis fan-out to `/linkedin:report`** (one analysis pipeline, not two). +- **Overlap with siblings** — `report` (which it now calls rather than re-implements). Subsumption test: `import` uniquely owns *data ingestion* (CSV detection, copy-to-exports, CLI invocation, baseline creation); `report` owns *presentation* of already-imported data. The Step 6 hand-off is the explicit de-duplication ("keeping a second analysis pipeline here drifted out of sync"). Clean separation — no subsumption, and a redundancy already removed. +- **Invocation leverage** — Moderate; "import analytics", "import CSV", "parse LinkedIn data". A distinct mechanical intent (you must import before you can report). +- **Recommendation** — **keep.** +- **Rationale** — Unique ingestion surface; the analysis overlap with `report` was already collapsed (delegation), which is exactly the discipline this audit endorses. + +### `report` + +- **Purpose** — Generate weekly/monthly/heatmap performance reports from imported data via the `trends` CLI: metrics table, top performers, 4-week trend, alert detection, `analytics-interpreter` (report mode) recommendations, markdown export. +- **Overlap with siblings** — `import` (which delegates *to* it), `analyze` (troubleshooting), `audit` (strategy review). Subsumption test: `report` is the only *recurring-numbers reporting* surface and the shared analysis engine `import` reuses. It does not diagnose recovery (`analyze`) or run the quarterly strategy lookback (`audit`). No subsumption. +- **Invocation leverage** — High and distinct: "weekly report", "performance report", "how did I do", "show my stats". Recurring intent. +- **Recommendation** — **keep.** +- **Rationale** — Canonical reporting engine, now the single analysis pipeline both it and `import` use; distinct from `analyze`/`audit`. + +### `ab-test` + +- **Purpose** — Full A/B testing lifecycle (design → log → analyze → history → suggest) with a directional-not-significant statistical honesty layer; delegates optimized-variant rewrites to `content-optimizer`. +- **Overlap with siblings** — `report`/`analyze` (it cross-references analytics data; offers "View weekly performance report"). Subsumption test: `ab-test` uniquely owns the *experiment* surface (hypothesis, variant design, running comparison, verdict heuristic). No other command designs or tracks experiments; it merely *reads* analytics for cross-reference. No subsumption. +- **Invocation leverage** — Moderate-to-high; specific trigger set ("A/B test", "test my hooks", "compare formats", "split test", "which hook works"). Distinct, directly-typed intent. +- **Recommendation** — **keep.** +- **Rationale** — Entirely unique experimentation surface; the analytics commands it links to don't touch experiments. + +--- + +## Group G — Growth / strategy (2) + +### `strategy` + +- **Purpose** — Phase-based growth plan (Phase 0–4 by follower count) + trajectory overlay + authority-building/signature-content compounding (Phase 2+). The canonical trajectory source `audit` defers to. Absorbed the former `authority` command (v2.0.0). +- **Overlap with siblings** — `audit` (names the gap, routes here for the fix), `profile` (which `strategy` defers to for profile-alignment), `monetize` (Phase 3+ monetization setup is mentioned but not owned here). Subsumption test: `strategy` uniquely owns the phase roadmap + trajectory prescription + authority compounding. It explicitly delegates profile to `profile`. No subsumption. +- **Invocation leverage** — High; very broad trigger set ("growth plan", "build authority", "signature content", "linkedin roadmap", "my best content"). Also the router's destination for milestone/follower-progress and the below-1K nudge target. +- **Recommendation** — **keep.** +- **Rationale** — Deep, canonical growth+authority surface that absorbed `authority`; the hub other commands (`audit`, the 1K-gated ones) route toward. + +### `profile` + +- **Purpose** — The canonical profile/topic-relevance optimization checklist (7 sections: Headline/About/Experience/Featured/Skills/Network/Engagement) for the 2026 relevance model + the profile-SEO/search-surface layer. The single source `analyze`, `audit`, and `strategy` all defer to. +- **Overlap with siblings** — `analyze` (Step 6 "If Profile-Content Mismatch" routes here), `audit` (Step 6 routes here), `strategy` (authority audit defers profile signals here). Subsumption test: `profile` is explicitly the *canonical* profile audit — three siblings delegate to it rather than duplicate it. None subsumes it; it subsumes none of them (no content/analytics/growth surface). This is textbook clean delegation. +- **Invocation leverage** — High and specific: "optimize profile", "profile audit", "fix my profile", "why is my reach low" (profile angle). Distinct, directly-typed intent. +- **Recommendation** — **keep.** +- **Rationale** — The single profile source-of-truth that three other commands point to; cutting/merging it would scatter the checklist they deliberately centralized. + +--- + +## Group H — Gated at ~1K followers (3) + +### `monetize` + +- **Purpose** — Monetization strategy: readiness scorecard, 4 stage-specific plans, lead-magnet blueprint, 4-week funnel, DM conversion workflow, CTA optimization, featured-section + revenue-model worksheets. Soft-gated at ~1K. +- **Overlap with siblings** — `strategy` (growth phases overlap the monetization stages; `strategy` Phase 3 mentions "lead magnets and monetization setup"), `outreach` (both 1K-gated revenue/opportunity surfaces). Subsumption test: `monetize` uniquely owns the *revenue mechanics* (pricing, offers, funnels, lead magnets, DM conversion). `strategy` only *mentions* monetization as a Phase 3+ focus area; it does not build a funnel or price an offer. `outreach` is collaborations/speaking, not products/funnels. No subsumption. +- **Invocation leverage** — Moderate; specific ("monetize", "lead generation", "consulting pipeline", "pricing strategy", "lead magnet"). Soft-gated, so naturally lower-frequency until the audience exists. +- **Recommendation** — **keep.** +- **Rationale** — Deep, unique revenue-mechanics surface; `strategy`'s monetization mention is a pointer, not a duplicate. The soft-gating is honest (works at any count, value compounds at 1K). + +### `outreach` + +- **Purpose** — Two-track orchestrator (collaborations + speaking) under one pitch paradigm: readiness, partner/event search + scoring, 12 collab formats + 4 talk templates, outreach messages, co-creation workflow + speaker portfolio, pipeline trackers (now state-persisted via `recordOutreachContact`), progression ladders. Absorbed the former `collab` + `speaking` commands (v2.0.0). Delegates to `network-builder`. +- **Overlap with siblings** — `monetize` (both 1K-gated), `strategy` (Phase 2+ collaborations). Subsumption test: `outreach` uniquely owns the partner/event pitch machinery. `strategy` only *names* collaborations as a focus area; `monetize` is products not partnerships. No subsumption. Internally it consolidated two commands and has an explicit capability checklist proving nothing was lost in that merge — a model of disciplined consolidation. +- **Invocation leverage** — Moderate; very broad trigger set spanning both tracks ("collaboration", "co-author", "speaking", "CFP", "pitch a talk", "outreach"). Soft-gated. +- **Recommendation** — **keep.** +- **Rationale** — Unique, deep two-track outreach surface that already consolidated `collab`+`speaking` cleanly; not subsumed. + +### `competitive` + +- **Purpose** — Competitive analysis of niche thought leaders: identify competitors (WebSearch), content/hook/engagement analysis, landscape map, gap analysis, differentiation strategy, inspired-not-copied takeaways. +- **Overlap with siblings** — `strategy` (differentiation is a growth lever there), `audit` (analyzes *your* content; this analyzes *others'*). Subsumption test: `competitive` is the only *external/competitor-facing* analysis surface. `audit` is inward-facing; `strategy` prescribes *your* plan, not a competitor scan. No subsumption. +- **Invocation leverage** — Low-to-moderate; specific ("competitive analysis", "analyze competitor", "what are others doing", "learn from others"). A periodic, distinct intent. CLAUDE.md/README describe it as 1K-gated in the same breath as monetize/outreach, though the file itself carries no gating logic. +- **Recommendation** — **keep** (with a minor honesty note). +- **Rationale** — Unique outward-facing competitive surface; nothing else scans competitors. Note: the router does not actually list `competitive` under the 1K-gated group nor apply the below-1K nudge to it (unlike `monetize`/`outreach`), and the command body has no follower check — a small documentation-vs-behavior inconsistency for the operator to reconcile, not a reason to merge or cut. + +--- + +## Summary table + +| Command | Recommendation | Merge target | One-line reason | +|---------|---------------|--------------|-----------------| +| `linkedin` | keep | — | Structural router; highest-leverage front door (minor map-staleness nit). | +| `post` | keep | — | Canonical long-band creation + refine/optimizer surface; core entry. | +| `quick` | keep | — | Unique short-band + the 8-template library `templates` folded into. | +| `react` | keep | — | Unique URL-ingestion + multi-source synthesis; directly-typed intent. | +| `pipeline` | keep (flag) | — | No clean subsumption, but a thin re-statement of `post`+`calendar`+`firsthour`; watch for erosion. | +| `carousel` | keep | — | Deep unique carousel/document engine + image generation. | +| `video` | keep | — | Unique LinkedIn-native video scripting + `video-scripter`. | +| `multiplatform` | keep (flag) | — | Unique Twitter/X + cross-platform surface; thinnest command, develop-candidate (could wire `content-repurposer`). | +| `newsletter` | keep | — | The long-form spine; everything long-form routes here. | +| `headless-review` | keep | — | Unique fresh-session cold-isolation a `newsletter` phase can't provide. | +| `pivot` | keep | — | Unique re-open state surgery `newsletter` delegates out. | +| `onboarding` | keep | — | Unique cohesive wizard + cross-phase triage; orchestrates, not duplicates. | +| `setup` | keep | — | Canonical full 8-category personalization; `onboarding` defers here. | +| `first-post` | keep | — | Unique first-timer scaffolding + `first_post_date` side-effect. | +| `batch` | keep | — | Unique multi-post-week + iCal; queue-write is shared plumbing. | +| `calendar` | keep | — | Canonical queue + publish action (absorbed `publish`). | +| `firsthour` | keep (note) | — | Unique worked first-hour plan; wiring point for orphaned `engagement-coach`. | +| `analyze` | keep | — | Unique troubleshooting/recovery surface; works without imported data. | +| `audit` | keep | — | Unique holistic strategy review; delegates fix to `strategy`/`profile`. | +| `import` | keep | — | Unique ingestion surface; analysis already delegated to `report`. | +| `report` | keep | — | Canonical reporting engine + shared analysis pipeline. | +| `ab-test` | keep | — | Entirely unique experimentation lifecycle. | +| `strategy` | keep | — | Canonical phase+trajectory+authority hub (absorbed `authority`). | +| `profile` | keep | — | Single profile source-of-truth three commands delegate to. | +| `monetize` | keep | — | Unique revenue-mechanics surface; honest soft-gating. | +| `outreach` | keep | — | Unique two-track pitch surface (absorbed `collab`+`speaking`). | +| `competitive` | keep (note) | — | Unique outward competitor scan; minor gating doc-vs-behavior nit. | + +**Tally: keep 27 · merge 0 · cut 0.** Two keeps carry a flag — `pipeline` (inconclusive, retained) and `multiplatform` (develop-candidate). + +--- + +## Count-impact summary (if every recommendation were applied as-is) + +Every recommendation is **keep**. Therefore, if all were applied: + +- **Commands removed:** 0 (0 cut + 0 merged-away). +- **Resulting count:** **27** — unchanged. +- **Lint / roster surfaces needing update:** **none.** `EXPECT_COMMANDS=27` in + `scripts/test-runner.sh` stays as-is; the CLAUDE.md command table (27), the + README command table, the `commands` header the lint checks, the router + `commands/linkedin.md`, and any SKILL roster all remain at 27. No roster + touched, no count contract changed. + +This mirrors the agent-overlap outcome in `overlap-measurement.md` (no agent +trimmed → 19/27 baseline unchanged): the command surface, audited under the same +subsumption discipline, also holds — every command has a unique surface or a +justified redundancy, and the two prior consolidations (`templates`/`publish`/ +`authority`/`collab`/`speaking` → `quick`/`calendar`/`strategy`/`outreach`, and +the `import`→`report` analysis delegation) have already removed the genuine +duplications. The surface is over-grown only in *count*, not in *redundancy*. + +The recommendations are informational; the operator decides per command. Two +non-blocking honesty nits surfaced along the way (each a wiring/doc fix, not a +merge/cut; a claimed third was withdrawn on verification): + +1. The router lists `post-feedback-monitor` (agent) under Post-Publish but not the + new `/linkedin:firsthour` command. +2. `calendar`'s publish-action first-hour block and `firsthour` share trigger + phrases ("just published") and should cross-link. +3. ~~`competitive` 1K-gating doc-vs-behavior.~~ **Withdrawn on verification:** + CLAUDE.md `:64`, README `:222`, the router, and the command body all leave + `competitive` ungated — only `monetize`/`outreach` carry "(unlocks at ~1K)". + No inconsistency exists; the claimed nit was unfounded. + +--- + +## Honest-uncertainty list (inconclusive — retain pending operator view) + +- **`pipeline`** — *inconclusive, retained.* It has no clean subsumption (it + uniquely *sequences* draft → schedule → first-hour → analysis into one guided + run), which under the trim rule mandates retention. But its constituent steps + are each a thinner inline condensation of richer leaf commands + (`post` / `calendar` / `firsthour`), and its trigger set is narrow/jargon-y, so + its real-world leverage as a *natural type* is the weakest in Group B. I could + not decide whether the orchestration value justifies the standalone command + versus folding its unique sequencing into `post` (with an explicit + schedule + first-hour hand-off). Per the methodology: **inconclusive — retain + pending operator view.** This is the one command where a future, operator-blessed + merge could be defensible; the evidence informs but does not compel it. + +All other 26 commands are clear **keep** decisions with no residual uncertainty. diff --git a/plugins/linkedin-studio/docs/remediation/finish-plan.md b/plugins/linkedin-studio/docs/remediation/finish-plan.md index a32471e..d04fd55 100644 --- a/plugins/linkedin-studio/docs/remediation/finish-plan.md +++ b/plugins/linkedin-studio/docs/remediation/finish-plan.md @@ -47,6 +47,21 @@ S17 (C13–C46 triage) → ALLOW [process complete] ## S14 — Command rationalization (re-opens the original command-surface Non-Goal) *Analysis → operator decision → execute. Nothing deleted without explicit per-command yes.* +> **AMENDMENT — S14 reframed by operator decision (2026-05-30): merge/cut → journey layer.** +> 14a's cold per-command review (`command-rationalization.md`) found **zero redundancy** — +> no command is a defensible merge/cut candidate (the two prior consolidations already +> removed the genuine overlaps). So instead of cutting, the operator chose to **add a +> journey layer over the kept atomics**. The build contract is +> **`journey-layer-design.md`** (this supersedes the 14b/14c "execute merges/cuts" bullets +> below — there are none to execute). **Delivered:** two new guided front-doors +> (`/linkedin:create`, `/linkedin:measure`) + the router re-tiered into five journeys +> (Start · Create · Engage · Measure · Grow) + `onboarding`/`strategy` elevated as the +> Start/Grow front-doors; the 27 atomic commands kept; **27 → 29 commands; v4.0.0 → v4.1.0 +> (minor/additive)**. The two 14a honesty nits real on verification were fixed (router lists +> `firsthour`; `calendar` cross-links to it); a third (a `competitive` 1K-gating claim) was +> withdrawn as unfounded. The **gate is unchanged**: `test-runner.sh` + `node --test` green +> → `/trekreview` **ALLOW** (no WARN-override) → commit own files → push. + - **14a Analysis (no edits):** cold per-command review of all 27 → `docs/remediation/command-rationalization.md`. Per command: purpose · overlap with siblings · invocation leverage (algorithmic + likely use) · recommendation **keep / develop / merge→X / cut** + rationale. Delegate the cold read to an diff --git a/plugins/linkedin-studio/docs/remediation/journey-layer-design.md b/plugins/linkedin-studio/docs/remediation/journey-layer-design.md new file mode 100644 index 0000000..17361d8 --- /dev/null +++ b/plugins/linkedin-studio/docs/remediation/journey-layer-design.md @@ -0,0 +1,154 @@ +# LinkedIn Studio — Journey-Layer Design (S14) + +> **Supersedes the S14 "merge/cut" framing.** 14a's cold review found 27 commands +> with **zero redundancy** (subsumption fails pairwise) — the surface is over-grown +> in *count*, not in *redundancy*. The operator therefore reframed S14 from +> "rationalize by cutting" to "**add a journey layer over the kept atomics**." This +> doc is the build contract. It also **absorbs S15's router-tiering** (finish-plan +> B2) — the router is re-tiered here, journey-style, so S15 collapses into S14. + +## Decisions locked (operator, 2026-05-30) + +1. **Architecture = journey layer.** Keep all 27 atomic commands as the *execution + tier*; add journey *front-door* orchestrators as the *primary surface*; subagents + remain the engine (no Workflow-tool / deep-research adoption — both are wrong-fit + for an everyday product surface, see the session reasoning). +2. **Realization = 2 new + 2 elevated.** Add `create` + `measure` as new front-doors; + elevate existing `onboarding` (Start) and `strategy` (Grow); **Engage** stays a + router-tier relay (`calendar` + `firsthour`), no front-door. Atomics → "advanced" + tier under each journey. +3. **Version = v4.1.0 (minor), NOT v5.0.0.** Honest SemVer: this is additive (2 new + commands) + presentation (router retier) + one tiny doc-vs-behavior reconcile + (`competitive`). No removal, rename, or behavior break on existing commands. + Precedent: v3.1.0 added 3 agents as a MINOR with a "reload required" note — same + shape. Reload note applies (the 2 new commands register at session reload). + +## Journey taxonomy (5 journeys + router) + +`Improve` is **not** a separate journey: improving is the action *after* measuring +(audit→fix, ab-test, `content-optimizer`), not a distinct entry. The operator's +"Improve + Analyze" is therefore folded into **Measure** (the measure→optimize loop). + +| Journey | Front-door | Atomics (advanced tier) | +|---------|-----------|--------------------------| +| **Start** | `onboarding` *(elevate — exists)* | setup · first-post | +| **Create** | **`create`** *(NEW)* | post · quick · react · carousel · video · multiplatform · batch · newsletter · pivot · headless-review · pipeline | +| **Engage** | *(router-tier — relay, no front-door)* | calendar · firsthour | +| **Measure** | **`measure`** *(NEW — `analyze` name is taken, so no collision)* | import · report · analyze · audit · ab-test | +| **Grow** | `strategy` *(elevate — exists)* | profile · competitive · monetize ⚿ · outreach ⚿ | +| *(map)* | `linkedin` router | — | + +⚿ = unlocks/soft-gates at ~1K followers. Count: Start 3 + Create 11 + Engage 2 + +Measure 5 + Grow 5 + router 1 = **27 atomics** + **2 new front-doors** = **29**. + +## The two new front-door commands + +Both are **thin guided routers** — they add *navigation* (a single "what do you +want?" entry), not duplicated logic. They **delegate** to the atomic that owns the +work. No drafting/analysis logic is copied into them. + +### `commands/create.md` — Content-creation front-door +- **Purpose:** one guided entry for "make something" when the user hasn't already + named the format. +- **Flow:** `AskUserQuestion` "What do you want to create?" → + - Short substantial post → `post` + - Quick 5-min post → `quick` + - React to a URL / article / news → `react` + - Carousel / document → `carousel` + - Video script → `video` + - Adapt for another platform → `multiplatform` + - A full week (batch) → `batch` + - Long-form / newsletter / essay → `newsletter` + - Then hand off to the chosen command (it owns the work). +- **Triggers (undirected create intent only — must NOT steal `post`/`quick` direct + intent):** "create", "make something", "create content", "what should I make", + "new content", "help me create", "linkedin create". +- **allowed-tools:** Read, Glob, AskUserQuestion. **No Write** (it delegates). + +### `commands/measure.md` — Performance/insight front-door +- **Purpose:** one guided entry for "understand how I'm doing." +- **Flow:** `AskUserQuestion` "What do you want to do?" → + - Import a new analytics CSV → `import` + - Weekly / monthly report → `report` + - Diagnose a problem (reach dropped, low engagement) → `analyze` + - Quarterly strategy audit → `audit` + - Design / review an A/B test → `ab-test` + - Then hand off to the chosen command. +- **Triggers (undirected measure intent):** "measure", "how am I doing", "my + performance", "show my analytics", "performance overview", "linkedin measure". +- **allowed-tools:** Read, Glob, AskUserQuestion. **No Write** (it delegates). + +## Elevation (2) — presentation only, no behavior change +- **`onboarding` = Start front-door.** Already the cohesive wizard. Router presents + it as the Start journey entry; body unchanged (optional one-line "Start journey" + framing only). +- **`strategy` = Grow front-door.** Already the phase/trajectory/authority hub. + Router presents it as the Grow journey entry; body unchanged. + +## Router re-tiering (`commands/linkedin.md`) — absorbs S15 B2 +Restructure the menu into the 5 journeys, each headed by its front-door with the +atomics listed beneath as "also / advanced": +- **Start** → `onboarding` · setup · first-post +- **Create** → `create` · post · quick · react · carousel · video · multiplatform · batch · newsletter · pivot · headless-review · pipeline +- **Engage** → calendar · firsthour *(tier, no front-door)* +- **Measure** → `measure` · import · report · analyze · audit · ab-test +- **Grow** → `strategy` · profile · competitive · monetize ⚿ · outreach ⚿ +- Keep the status line (weekly progress / streak / follower phase) + queue summary. +- **Fixes nit #1:** `/linkedin:firsthour` is now listed (under Engage) — the router + previously listed only the `post-feedback-monitor` agent there. + +## Honesty nits fixed in lockstep (from 14a) +1. **Router omits `firsthour`** → fixed by the retier (Engage tier lists it). +2. **`calendar` ↔ `firsthour` trigger overlap** → `calendar.md`'s publish-action + first-hour block gets a one-line cross-link: "for the full worked sprint plan, + run `/linkedin:firsthour`." +3. **`competitive` gating doc-vs-behavior** → **14a's nit was unfounded** (verified + against the files: CLAUDE.md `:64`, README `:222`, the router, and the command body + all leave `competitive` **ungated** — only `monetize`/`outreach` carry "(unlocks at + ~1K)"). There is no inconsistency to fix. Competitive analysis helps at any stage + (especially early positioning), so it stays ungated. Action: place it in the Grow + tier marked **"Any phase"** (no ⚿), and add a one-line note to the router's gating + rule that `competitive` is explicitly not gated. (An interim router edit that + mistakenly added a ~1K marker was reverted in the same pass.) + +## 14a doc correction (`command-rationalization.md`) +- Add the missing **`multiplatform`** entry: **keep** (unique Twitter/X-thread + + cross-platform surface; subsumption fails as a cut), flagged **thinnest / + develop-candidate** (only command that inlines static templates instead of + delegating — could use `content-repurposer`). Belongs in **Create**. +- Fix the two header-count typos (Group C "(4)"→3; Group F "(4)"→5). +- Corrected tally: **keep 27 (27/27 covered)** · 0 merge · 0 cut · 1 develop-candidate. + +## Count / version / roster lockstep (the release mechanics) +- `scripts/test-runner.sh`: `EXPECT_COMMANDS` **27 → 29**. +- Version **4.0.0 → 4.1.0** everywhere it appears (version-sync rule: grep both + `4.0.0` and the `27`-count claims). +- Three-doc rule (same commit / immediately after): + - `CLAUDE.md` (plugin): command table +2 rows, count 27→29, version, v4.1.0 para. + - `README.md` (plugin): command table, count, version, CHANGELOG entry. + - Root `README.md` (marketplace): the linkedin-studio line. +- Also: `CHANGELOG.md` if present; any SKILL roster enumerating commands; the router + roster (done above); `STATE.md` count block (27→29, version 4.1.0). + +## Verification (testable — gate before push) +- `ls commands/*.md | wc -l` == **29**. +- `bash scripts/test-runner.sh` exit **0** (EXPECT_COMMANDS=29; stat/version/model + consistency + render-chain + $-safety guards all green). +- `node --test hooks/scripts/__tests__/*.test.mjs` — all pass (no hook logic touched + → expect the existing **98**). +- `grep -rn "4\.0\.0\|27 command" commands README.md CLAUDE.md` (user-facing) → 0 + stale hits (historical CHANGELOG mentions allowed). +- Router renders 5 journeys; primary tier = the 5 front-doors (`create`, `measure`, + `onboarding`, `strategy` + Engage tier); atomics nested; 1K-gated flagged; + `firsthour` present. +- `create` / `measure`: valid frontmatter, `AskUserQuestion` routing, **delegate + only** (grep shows no inlined drafting/analysis logic, no `Write` in create/measure). +- `/trekreview --project docs/remediation/` → **ALLOW** (no WARN-override) → commit + (own files only) → push origin. + +## Out of scope (this session / S14) +- Aggressive consolidation / folding atomics (operator chose keep-all). +- Building the `multiplatform` "develop" (content-repurposer wiring) — recorded as a + keep + develop-candidate, not built unless requested. +- S16 (saves) / S17 (C13–C46) — later finish-plan steps. +- UI-brief M0 (move mutable data out of tree) — pending; conflicts with S16, not S14. diff --git a/plugins/linkedin-studio/docs/remediation/review.md b/plugins/linkedin-studio/docs/remediation/review.md index 6246c61..d8c000a 100644 --- a/plugins/linkedin-studio/docs/remediation/review.md +++ b/plugins/linkedin-studio/docs/remediation/review.md @@ -1,13 +1,13 @@ --- type: trekreview review_version: "1.0" -task: "Remediate linkedin-studio from the baseline audit — correctness, honesty, generalization, and the highest-leverage 2026 coverage gaps (full Phase 0–3 roadmap, phased)" +task: "S14 — journey layer over the LinkedIn Studio command surface (operator-reframed from merge/cut after 14a found zero redundancy)" slug: remediation project_dir: docs/remediation/ brief_path: docs/remediation/brief.md -scope_sha_start: c5b4c58f4f390aca83c8937880c5fd0bcc983e44 -scope_sha_end: 36f79dd702b9315a0cd9100c3a8dd6dd81b3797f -reviewed_files_count: 50 +scope_sha_start: 431a893 +scope_sha_end: 431a893 +reviewed_files_count: 15 verdict: ALLOW mode: default effort: high @@ -15,152 +15,132 @@ profile: premium findings: [] --- -# Review — linkedin-studio audit-remediation (S13 re-review: `$`-class closure + scalar-test fix) +# Review — linkedin-studio S14 (journey layer; merge/cut → add a layer) ## Executive Summary -**Verdict: ALLOW** — 0 BLOCKER, 0 MAJOR, 0 MINOR, 0 SUGGESTION. +**Verdict: ALLOW** (post-remediation) — 0 BLOCKER, 0 MAJOR, 0 MINOR, 0 SUGGESTION open. +Two independent reviewers (brief-conformance, code-correctness) ran COLD, high-effort, +without cross-feeding, on the as-delivered uncommitted working tree (HEAD `431a893` + +the S14 delta). **Each returned 2 MAJOR findings; deduped to 3 distinct issues.** All +three were **remediated in-session** with deterministic re-verification; the final +state that pushes is clean. -This is the **S13 re-review** — the **seventh** full-brief sweep -(`c5b4c58..36f79dd` + the uncommitted S13 working-tree delta), run COLD and -high-effort. S13 was commissioned to close the two findings the S12 re-review -left open (verdict WARN, 0/1/1/0): the MAJOR `MISSING_TEST` (the S12 `$`-bearing -test asserted the Recent Posts section but never the `last_post_topic` scalar, so -the corruption shipped green) and the MINOR `MISSING_ERROR_HANDLING` (the -`last_post_topic` scalar was still `$`-unsafe because `replaceField` used a -replacement *string* for untrusted content). Two independent reviewers -(brief-conformance, code-correctness) ran without cross-feeding; the coordinator -applied bounded dedup + the HubSpot Judge filters + verdict (high-effort → -Cloudflare reasonableness filter skipped; the operator weighs borderline findings). -**Both reviewers returned empty finding sets.** +S14 was **reframed by operator decision (2026-05-30)**: 14a's cold per-command review +(`command-rationalization.md`) found **zero redundancy** across the 27 commands (no +defensible merge/cut), so instead of cutting, a **journey layer** was added over the +kept atomics. Build contract: `journey-layer-design.md`. Delivered: two new +delegate-only front-doors (`/linkedin:create`, `/linkedin:measure`), the router +re-tiered into five journeys (Start · Create · Engage · Measure · Grow), `onboarding`/ +`strategy` elevated as the Start/Grow front-doors; 27 → 29 commands; v4.0.0 → v4.1.0 +(minor/additive). Two of 14a's honesty nits (router lists `firsthour`; `calendar` +cross-links to it) were real and fixed; a third (a `competitive` 1K-gating claim) was +withdrawn as unfounded on verification. -**Both S12 findings are CLOSED at the reviewed (uncommitted) state:** -- `replaceField` (`hooks/scripts/state-updater.mjs:14-24`) now uses a replacement - **function** (`() => \`${field}: ${value}\``), so the untrusted `last_post_topic` - at the `:64`-equivalent call site is inserted verbatim — no `$&`/`$1`/`` $` `` - expansion. The MINOR is closed. -- The existing `$`-bearing test (`hooks/scripts/__tests__/state-updater.test.mjs`) - now carries `assert.match(result.content, /^last_post_topic: "\$100 budget — \$& - and \$1 rule"$/m, …)`, distinct from the section-entry `includes()` it already - had. This assertion **fails on the old string-`replaceField` and passes on the - function form** (orchestrator-verified by reverting the fix: the test went - 40 pass / 1 fail). The false-green MAJOR is closed. - -**The class — not just the line — is closed.** The recurring S9→S12 lesson is -"close the class, not the line"; the class here is "untrusted user content reaching -ANY `String.replace` replacement *string*". Beyond the `replaceField` scalar, S13 -also converted the three remaining additive-insert sites (`recordFirstHourPlan` -`:246`-equiv; `recordOutreachContact` `:305/:308`-equiv) from a string replacement -carrying an intentional `$1` backref + interpolated date to a replacement function -(`(m) => \`${m}\n…\``). The code-correctness reviewer verified rigorously that this -is **behavior-preserving**: each regex's capture group spans the *entire* match (the -only chars outside the group are the zero-width `^`/`$` anchors), so the full match -`m` is character-identical to the old `$1`. After S13, **every `.replace()` in -`state-updater.mjs` uses a replacement function or a `$`-free literal** — the class -is closed by construction, not by per-line patch. - -**A structural guard replaces the per-line proof.** New -`scripts/check-replace-safety.mjs` (wired as `test-runner.sh` Section 12) proves the -property behaviorally: it drives every exported mutator with an adversarial payload -of every special replacement token (`$&`, `` $` ``, `$'`, `$$`, `$n`) in every -free-text *and* date field and asserts the payload survives verbatim. Two structural -backstops run on every invocation — **coverage-completeness** (a newly-exported -mutator without `$`-coverage fails the guard) and a **non-vacuity self-test** (a -naive string-replace MUST corrupt the payload and a function MUST preserve it, else a -PASS is meaningless), mirroring the Section 8/10/11 self-tests. The orchestrator -mutation-proved it end-to-end: reverting `replaceField` to a string makes the guard -exit 1 with two findings; restoring it returns exit 0. - -**No Phase-0–3 Success Criterion regressed.** The brief-conformance reviewer traced -each S13 clause to delivered code and confirmed the counts (19 agents / 27 commands / -25 references / 6 skills), the version (4.0.0), the single-source algorithm-signal, -the model-consistency guard, and the render-chain-propagation guard all still hold; -S13 touched no command/agent/reference file. The two Non-Goals the brief amendment -re-opens (the command invocation surface for S14; saves manual-entry for S16) trace -to **explicit operator decisions** in the brief amendment, and S13 itself did not -touch either surface — no `SCOPE_CREEP_BUILT`. - -**Push decision: ALLOW.** The two S12 findings are closed, the class is closed -structurally, the lint is non-vacuous and mutation-proven, all suites are green -(`scripts/test-runner.sh` → 71/0/0; `node --test` → 98/98), and no SC regressed. The -ORIGINAL remediation brief now closes clean. Per `feedback_trekreview_always_last` + -Handover 6, this review is the gate; with ALLOW, S13 may push. +**What both reviewers confirmed conformant + correct** (the bulk of the delivery): +the two front-doors are delegate-only (no `Write`; route to the contracted targets); +the router re-tier preserves **reachability for all 29 commands** (no atomic dropped; +`post-feedback-monitor` still mentioned in prose, not orphaned); `EXPECT_COMMANDS` +27→29 + `## Commands (29)` + `ls`=29 agree; version 4.1.0 is consistent across +plugin.json / README version badge / CLAUDE.md header / CHANGELOG / root docs; SemVer +framing is honest (additive, nothing falsely "breaking", nothing removed/renamed); +out-of-scope discipline held (no atomic folded; `multiplatform`-develop only recorded, +not built; S16/S17/UI-brief-M0 untouched). ## Coverage -Scope SHA range: `c5b4c58` (= `origin/main`, parent of remediation Steg 1) → -`36f79dd` (HEAD, the S12 commit) **plus the uncommitted S13 working-tree delta** +Scope: HEAD `431a893` (S13's commit) + the **uncommitted S14 working-tree delta** (annotated `[uncommitted]` — a brief-level contract; the brief's Assumptions allow -uncommitted review). The committed range (47 files) was already deep-reviewed and -cleared at S12 except the 2 WARN findings; the active S13 delta is the 9 working-tree -files below. **No silent skips.** +uncommitted review). 15 files = the operator's own changes; the 3 untracked +not-mine files (`docs/linkedin-studio-persona-brief.md`, `…-ui-brief.md`, +`docs/voyage-build/progress.json`) are explicitly excluded from scope and from the +commit. **No silent skips.** | Treatment | Count | Notes | |-----------|-------|-------| -| `deep-review` (hooks/** + the new guard) | 4 | `state-updater.mjs`, `state-updater.test.mjs` `[uncommitted]`; `scripts/check-replace-safety.mjs` `[uncommitted, new]`; `scripts/test-runner.sh` `[uncommitted]` | -| `summary-only` | 46 | the committed `c5b4c58..36f79dd` range (already cleared at S12) + the S13 doc edits `CLAUDE.md`/`README.md`/`docs/integration-test-guide.md` `[uncommitted]` + `docs/remediation/{brief.md (amendment), finish-plan.md}` `[uncommitted]` | +| `deep-review` | 6 | `commands/create.md` `[new]`, `commands/measure.md` `[new]`, `commands/linkedin.md` (router re-tier), `commands/calendar.md`, `scripts/test-runner.sh`, `README.md` | +| `summary-only` | 9 | plugin `CLAUDE.md`/`CHANGELOG.md`/`plugin.json`/`commands/onboarding.md`; root `CLAUDE.md`/`README.md`; `docs/remediation/{finish-plan.md, command-rationalization.md, journey-layer-design.md}` | | `skip` | 0 | no lockfiles / svg / generated / dist | **Cross-cutting execution criteria (run by orchestrator):** -`scripts/test-runner.sh` → 71 passed / 0 failed / 0 warnings, exit 0 (was 70; -+1 — Section 12 `$`-safety guard). -`node --test hooks/scripts/__tests__/*.test.mjs` → 98 tests, 98 pass, 0 fail (the -S13 scalar assertion was added to an existing test, not a new test). -`node scripts/check-replace-safety.mjs` → exit 0 (8 adversarial cases / 5 mutators, -coverage-complete, self-test non-vacuous); mutation-proven (reverting the fix → exit 1). - -**S12 findings — both confirmed CLOSED:** -`replaceField` (`state-updater.mjs:14-24`) → replacement function; the `$`-bearing -test now pins the `last_post_topic` scalar (`state-updater.test.mjs`); the three -remaining additive-insert string-replacements (`:246/:305/:308`) → functions; the -class is closed and guarded structurally (Section 12). +`scripts/test-runner.sh` → **74 passed / 0 failed / 0 warnings**, exit 0 (was 71; ++2 the two new commands' frontmatter checks, +1 a new README commands-badge guard, +−1 net rounding of the count-loop — all green). `node --test …/*.test.mjs` → **98/98** +(no hook logic changed). `ls commands/*.md` → **29**. ## Findings -None. Both independent reviewers returned empty finding sets; the coordinator's -bounded passes (dedup, HubSpot Judge, verdict) on an empty set yield ALLOW. +**3 distinct MAJOR findings were raised by the independent reviewers and ALL closed +in-session.** No finding remains open; the trailing JSON `findings` is empty because +the pushed state carries none. + +### [MAJOR — RESOLVED] README commands badge stale at `commands-27` +*Raised by both reviewers (`COUNT_DRIFT` / `SUCCESS_CRITERION_UNMET`), `README.md:11`.* +The release reconciled the version badge, intro prose, CLAUDE.md header, `EXPECT_COMMANDS`, +and rosters to 29, but the shields **commands** badge still read `commands-27-green`. It +slipped both the human pass and the gate (the version-consistency grep checks only the +*version* badge; the design-doc verification grep searched `"27 command"` with a space, +which cannot match `commands-27-green`). +**Resolution:** `README.md:11` → `commands-29-green`; **and** a new +`test-runner.sh` Section-2 guard now asserts the `badge/commands-${EXPECT_COMMANDS}-` +shields badge matches the contract (closing the class, not just the line — the +count-badge analogue of the existing version-badge check). Re-verified: lint 74/0/0; +`grep commands-27` → 0 hits. + +### [MAJOR — RESOLVED] Router self-contradicted on `competitive` gating +*Raised by code-correctness (`INTERNAL_CONTRADICTION`), `commands/linkedin.md:140`.* +The Grow journey **table** marked `competitive` "Any phase" and the gating-rule note +said it is **not** gated, but the **"Ask the User"** menu line still rendered +`competitive ⚿` (the file's own ~1K soft-gate marker). An interim erroneous gating +(added then reverted in the same pass) had been reverted in the table + the gating note ++ the design doc, but the menu line was missed — the exact inconsistency the release +claims to fix. +**Resolution:** `commands/linkedin.md` Ask-the-User Grow line → `competitive` (no ⚿); +re-verified `grep "competitive ⚿"` → 0 hits across commands + the design doc. The router +now states competitive ungated consistently in the table, the note, and the menu. + +### [MAJOR — RESOLVED] STATE.md binding count block stale at 27 / v4.0.0 +*Raised by brief-conformance (`SUCCESS_CRITERION_UNMET`), `STATE.md:46`.* +The design-doc lockstep enumerates the STATE.md count block as a surface to update +(27→29, v4.1.0). `STATE.md` is **gitignored** (not part of the pushed commit, so not a +push-blocking three-doc violation), but it is a contracted lockstep item and would +mislead the next session. +**Resolution:** updated to 29 / v4.1.0 as part of the session-close STATE.md overwrite +(which also advances the pointer to S15). Not in the pushed artifact (gitignored). ## Remediation Summary -**Gate: ALLOW.** The S12 WARN is fully resolved: the `replaceField` scalar -`$`-corruption (MINOR) and the false-green test (MAJOR) are both closed; the -`$`-injection class is closed across the whole `state-updater.mjs` mutation surface; -and a behavioral, coverage-complete, self-testing Section-12 lint guards it -structurally against future regressions. All suites green; no Phase-0–3 SC regressed; -the two re-opened Non-Goals trace to explicit operator decisions and S13 stayed in its -lane. +**Gate: ALLOW (post-remediation).** Two independent high-effort reviewers found 3 +distinct MAJORs — all stale-count / incomplete-revert completion-misses of THIS +session's own committed lockstep, none a design flaw. All three were closed in-session +with deterministic re-verification (lint 74/0/0; targeted greps → 0 residual hits); a +new lint guard prevents the badge regression class from recurring. No BLOCKER; no +design or correctness defect in the journey layer itself; reachability for all 29 +commands preserved; SemVer honest. The pushed state is clean. -**Two non-blocking observations, recorded by both reviewers, neither rising to a -catalogue finding:** -1. The S13 dead binding both reviewers named (`check-replace-safety.mjs` `HERE` + - its now-unused `node:url`/`node:path` imports) was removed during this review - pass; the guard remains green (exit 0) after removal. -2. The behavioral guard catches a new *unguarded exported mutator* (coverage - backstop) but not a new unsafe `String.replace` added *inside* an existing - battery-covered mutator on a field the battery does not fuzz. This is a documented, - deliberate limit of the behavioral proxy (vs a per-`.replace()` AST enumeration) - and is **moot today** — every `.replace()` in `state-updater.mjs` is already a - function. Recorded as a known boundary, not a defect; closing it further is out of - S13 scope (no real `$`-unsafe site exists to catch). +Decision rationale for in-session fix (vs `feedback_trekreview_always_last`'s +fix-in-next-session): the three findings are mechanical completions of the current +task's own lockstep — #1 and #2 would have failed the design doc's own Verification +grep, #3 is an incomplete revert — i.e. *unfinished scope of S14*, not deliberation- +needing review findings. Shipping a README badge that reads "27" and a router that +self-contradicts on gating would be dishonest to the release's own claims. Closing them +now reaches a genuine ALLOW (not a WARN-override). -The two adjacent machine-value `.replace()` sites the correctness reviewer probed — -`session-start.mjs:396` (`${actualWeek}`, a computed ISO week) and `week-rollover.mjs` -(computed week / literal int) — carry no untrusted content and are therefore not -members of the defect class, consistent with how the S12 review itself classified the -date/integer `replaceField` call sites. No finding. - -Per Handover 6, this `review.md` is consumable by -`/trekplan --brief docs/remediation/review.md`; the trailing JSON block is the machine -contract for that handover. With an ALLOW verdict and no BLOCKER/MAJOR findings, no -follow-up remediation plan is required — the ORIGINAL brief is closed clean and the -finish-plan continues at S14. +Per Handover 6, this `review.md` is consumable by `/trekplan --brief …`; with an ALLOW +verdict and no open findings, no follow-up plan is required — S14 may commit + push, and +the finish-plan continues at S15. ```json { "verdict": "ALLOW", - "scope": { "sha_start": "c5b4c58f4f390aca83c8937880c5fd0bcc983e44", "sha_end": "36f79dd702b9315a0cd9100c3a8dd6dd81b3797f", "reviewed_files_count": 50, "uncommitted_delta": true }, + "scope": { "sha_start": "431a893", "sha_end": "431a893", "reviewed_files_count": 15, "uncommitted_delta": true }, "counts": { "BLOCKER": 0, "MAJOR": 0, "MINOR": 0, "SUGGESTION": 0 }, "findings": [], + "resolved_in_session": [ + { "severity": "MAJOR", "title": "README commands badge stale at 27", "file": "README.md", "line": 11, "rule_key": "COUNT_DRIFT", "resolution": "badge -> commands-29-green + new test-runner.sh commands-badge guard" }, + { "severity": "MAJOR", "title": "Router self-contradicts on competitive gating", "file": "commands/linkedin.md", "line": 140, "rule_key": "INTERNAL_CONTRADICTION", "resolution": "Ask-the-User Grow line competitive ⚿ -> competitive; 0 residual hits" }, + { "severity": "MAJOR", "title": "STATE.md binding count block stale at 27 / v4.0.0", "file": "STATE.md", "line": 46, "rule_key": "SUCCESS_CRITERION_UNMET", "resolution": "updated to 29 / v4.1.0 in session-close overwrite (gitignored, not pushed)" } + ], "dropped_findings": [] } ``` diff --git a/plugins/linkedin-studio/scripts/test-runner.sh b/plugins/linkedin-studio/scripts/test-runner.sh index 8ce02f7..3fd252d 100755 --- a/plugins/linkedin-studio/scripts/test-runner.sh +++ b/plugins/linkedin-studio/scripts/test-runner.sh @@ -42,7 +42,7 @@ warn() { echo -e "${YELLOW}⚠${NC} $1"; WARN=$((WARN + 1)); } # Source of truth: CLAUDE.md headers + STATE.md Telling. Bump these together # with the files when adding/removing an agent, command, reference, or skill. EXPECT_AGENTS=19 -EXPECT_COMMANDS=27 +EXPECT_COMMANDS=29 EXPECT_REFS=25 EXPECT_SKILLS=6 @@ -101,6 +101,17 @@ else fail "CLAUDE.md commands header ($DOC_COMMANDS) != contract ($EXPECT_COMMANDS)" fi +# README shields commands-count badge must match the contract too. Added after an +# S14 /trekreview found the badge stale at commands-27 while the surface shipped 29: +# the version-consistency grep (Section 9) checks only the version badge, and the +# count guards above check the CLAUDE.md header, so the README count badge slipped +# both. This closes that gap (the count-badge analogue of the version-badge check). +if grep -q "badge/commands-${EXPECT_COMMANDS}-" README.md; then + pass "README commands badge declares ${EXPECT_COMMANDS}" +else + fail "README commands badge != ${EXPECT_COMMANDS} (expected shields badge/commands-${EXPECT_COMMANDS}-)" +fi + echo "" # --- Section 3: Agent Frontmatter ---