14 tolerant parsers per kanonisk archetype-routing-tabell (Step 11) + 3 helpers (parseTable, parseSections, extractField). Each parser returns {ok:true, data} or {ok:false, errors:[{section, reason}]} — never throws on bad input. PARSERS routing-objekt eksponert via window.__PARSERS.
Verified against all 17 fixtures: every parser produces expected shape. Empty input returns structured error per Verify-asserts.
Synthetic markdown fixtures for the 17 report-producing commands per the canonical archetype-routing-tabell. Each fixture uses the consistent ANPR-trafikkanalyse system from brief example to produce parser-input that exercises every archetype path (aiact, requirements-list, text-document, fria, conformity-checklist, matrix-risk 5x5, matrix-risk-6x5, findings, cost-distribution, capability, phased-plan, markdown, verdict, comparison).
Real /architect:<command> capture deferred to incremental work; synthetic fixtures suffice as parser test input for Steps 11-12.
Step 9 of v3 plan. Replaces renderCatalogStub with full
renderCatalogSurface — search-input + 5 .expansion-grupper (en per
CATALOG.categories) + per-command-card with "Åpne skjema"-button. Klikk
åpner modal med renderCommandForm (samme generic renderer som
prosjekt-detalj fra Step 8).
Søk: input-event oppdaterer modul-lokal catalogSearchQuery og kaller
refreshCatalogResults() som re-rendrer kun groups-containeren — bevarer
fokus + cursor i søkefeltet (full re-render ville flyttet caret).
Filtrerer på id+label+description+argument_hint. Når query er aktiv
forces alle expansions med treff åpne; ellers er 'regulatory' åpen som
default (mest brukt entry-point).
Verktøy-commands får .catalog-card__pill="Verktøy" + .catalog-tool-notice
("Verktøy — ingen rapport-import"). Modalen viser samme advarsel via
.guide-panel--info-banner. Rapport-produserende får "Rapport"-pill.
Verifisert via vm-sandbox med activeSurface='catalog':
- data-command-card === 24 (Step 9 verify-assert ✓)
- 5 expansion-grupper (data-catalog-group)
- 24 open-catalog-form-knapper
- 17 Rapport-pills + 7 Verktøy-notices (matcher CATALOG.commands.filter
produces_report)
- refreshCatalogResults() med query='classify' kjører feilfritt
Step 8 of v3 plan. renderCommandForm(commandId, opts) reads
CATALOG[id].input_fields and emits a form with all 6 supported field types
(text/textarea/select/multiSelect/boolean/number). Shared fields
auto-prefill from state.shared via field.shared_path dot-lookup; local
fields prefill from project.reports[id].input when opts.projectId is set.
window.__buildCommand(commandId, formData) builds /architect:<id>
key="value" key="value" ... — shared fields merged first (CATALOG order),
formData overrides and may include keys outside the catalog (passthrough).
Empty/null/empty-array values omitted. Multi-values comma-joined inside
quotes; quotes/backslashes escaped.
Copy-button writes via navigator.clipboard.writeText with graceful
fallback to inline preview when clipboard is blocked (file:// in some
browsers). Preview-button shows the same string without copying.
Replaces the form-zone-placeholder in renderCommandSubCard. All 24
command-cards in project-detail now render real forms (verified:
data-command-card === 24, data-command-form === 24, copy-command
buttons === 24, field-from-tag === 39, paste-import === 17,
report-slot === 17, buildCommand('classify',{riskLevel:'høy'}) →
'/architect:classify organisation_name="Vegvesen" sector="Statlig"
riskLevel="høy"').
Step 6/17 av Playground v3-leveransen (Session 2, Wave 2).
Hjem-skjerm med 3-track entry-pattern (.tracks__card--guided/explore/expert):
- Onboard / Re-onboard
- Nytt prosjekt
- Command-katalog
Prosjekt-liste under tracks: .fleet-grid med .fleet-tile per prosjekt
(navn + scenario-chip + meter med rapport-fremdrift). Tom-state vises
som .guide-panel--info med 'Opprett første prosjekt'-knapp.
Topbar (renderTopbar) med brand + nav + eksport/import-knapper synlig
på home/catalog/project. Onboarding holdes uten topbar for full-fokus
første-flyt. import-input change-handler ruter via window.__importState
fra Step 3 og kjører scheduleRender etter import.
Verifisert via vm sandbox:
- 21 tracks__card-treff (3 cards med modifier-klasser)
- guided/explore/expert-modifiers alle til stede
- empty-state guide-panel--info når projects=[]
- fleet-grid suppressed når projects=[]
Stub-actions for new-project (Step 7 erstatter med modal-åpning).
README/CLAUDE.md-update deferred til Step 17 (Session 5).
Step 5/17 av Playground v3-leveransen (Session 2, Wave 2).
5 grouped sections (organization/technology/security/architecture/business)
rendered with Tier 3 .form-progress sidebar and .expansion components per
group. Validation via .error-summary with click-to-focus links.
ONBOARDING_SCHEMA mirrors agents/onboarding-agent.md Phase 1-5 (18 fields
total). commitOnboarding() writes to state.shared.<group>.<field> via
Proxy → throttled IDB/localStorage write. Re-onboard is just navigate
back to onboarding — pre-fills from state automatically.
Verified via vm sandbox: bootstrap auto-routes to onboarding when no
org.name, commitOnboarding produces >=5 keys in shared.organization,
validation catches required-empty (2) and accepts filled (0).
Surface routing: showSurface() toggles [hidden] across data-surface
sections. scheduleRender batches via queueMicrotask. Action router
dispatches data-action attributes to ACTIONS map. README/CLAUDE.md-update
deferred til Step 17 (Session 5).
Step 3/17 av Playground v3-leveransen.
Eksport:
- buildEnvelope(): { appId, schemaVersion, exportedAt, shared, projects,
activeProjectId, activeSurface, preferences } — JSON.parse(JSON.stringify(...))
for å strippe Proxy-wrappere
- exportState(): Blob + URL.createObjectURL + programmatisk <a download>-klikk
+ revokeObjectURL etter 0ms timeout. File System Access API krever HTTPS
(secure context) og er ikke tilgjengelig på file:// — derfor Blob-pattern.
- Filnavn-format: ms-ai-architect-playground-<ISO-stamp>.json
Import:
- importState(File): file.text() -> JSON.parse -> envelope-validering (appId
+ schemaVersion required) -> migrateState() -> persistence.save() -> in-place
state-update (Proxy-binding må bevares — kan ikke bytte raw-referansen)
-> manuell 'change'-event-dispatch så subscribers re-rendrer
- file.text() er Promise<string> som fungerer på file:// uten secure context
MIGRATIONS-pipeline:
- Eager: alle migrasjoner kjøres sekvensielt fra fil-versjon til SCHEMA_VERSION
ved import (ikke lazy ved access)
- Nøkkel-format: 'N->M' (fortløpende). Aldri hopp over et steg.
- Kaster eksplisitt feil ved manglende migrasjons-funksjon eller ved
funksjon som ikke setter schemaVersion korrekt — silent corruption
unngås (brief Risk High).
Eksponerte globals: __buildEnvelope, __exportState, __importState, __MIGRATIONS.
Verify-assert: JSON.parse(JSON.stringify(window.__buildEnvelope())).schemaVersion === 1
Plan: .claude/projects/2026-05-03-playground-v3-architecture/plan.md (Step 3)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Step 2/17 av Playground v3-leveransen.
State-skjelett:
- StateBus extends EventTarget (sharedBus + projectBus)
- Dyp Proxy med set/deleteProperty-traps som batcher dispatchEvent via
queueMicrotask (N synkrone mutasjoner -> én change-event per tick)
- Path tracking: subscribers får detail.paths for å filtrere relevante grener
- INITIAL_STATE med shared.{organization,technology,security,architecture,
business} + projects[] + activeProjectId/Surface + preferences.theme
Persistens:
- IDB primær: én DB ('ms-ai-architect-playground-v1') med 3 stores
(shared, projects, meta). Promise-wrapper rundt indexedDB.open.
- Synkrone migrasjoner i onupgradeneeded med oldVersion-guards (callback-stil
cursor — async cursor-iterasjon er forbudt per w3c/IndexedDB#282)
- db.onversionchange = () => db.close() defensivt på alle koblinger
- localStorage-fallback ved IDB-feil (Safari private mode, kvote): rå JSON
i STATE_KEY, warn ved >4.5 MB nær 5 MiB cap
- Throttled writer: debounce 300 ms etter siste mutasjon
Bootstrap:
- Auto-kjørt på slutten av <body> (DOM allerede parsed)
- window.__store + window.__persistence eksponert for Verify-asserts
Verify-asserts (i nettleser-konsoll på file://-åpnet HTML):
typeof window.__store !== 'undefined' && window.__store.state.schemaVersion === 1
Plan: .claude/projects/2026-05-03-playground-v3-architecture/plan.md (Step 2)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Step 1/17 av Playground v3-leveransen (Session 1, Wave 1).
- Single-file HTML med klassisk inline-script (file://-kompatibel per WHATWG
html#8121: external type=module-scripts feiler på file:// i Chrome+Firefox)
- 7 vendored CSS-link-tags i korrekt rekkefølge: fonts, tokens, base, components,
components-tier2, components-tier3, components-tier3-supplement
- 4 placeholder-overflater (#surface-onboarding, #surface-home, #surface-catalog,
#surface-project) — fylles ut i Steps 5-7
- IIFE med STATE_KEY ('ms-ai-architect-state-v1') og SCHEMA_VERSION (1) konstanter
- Eksponerer __STATE_KEY og __SCHEMA_VERSION på window for Verify-asserts
- v2-fila beholdes parallelt frem til Step 17 (sletting)
Plan: .claude/projects/2026-05-03-playground-v3-architecture/plan.md
Brief: .claude/projects/2026-05-03-playground-v3-architecture/brief.md
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Establish a single governance document at marketplace root and copy
it into each of the 9 plugins so every plugin folder remains 100%
self-contained. Replace the inconsistent provocative blurb across
all READMEs with a uniform fork-and-own paragraph that links to
the local GOVERNANCE.md.
[skip-docs]
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Renames playground/azure-ai-playground.html to
playground/ms-ai-architect-playground.html (history preserved via git mv).
Old name was too narrow — plugin covers the full Microsoft AI stack
(Foundry, Copilot Studio, M365 Copilot, Power Platform, Agent Framework).
Replaces the inline <style> block with seven <link> tags pointing at the
vendored design-system under playground/vendor/playground-design-system/:
fonts.css, tokens.css, base.css, components.css, components-tier2.css,
components-tier3.css, components-tier3-supplement.css.
A small inline shim maps legacy playground tokens (--bg, --surface,
--accent, --gradient1) onto design-system tokens (--color-bg,
--color-surface, --color-primary-500, etc.), keeping all existing
playground-specific class CSS (.hero, .wizard-card, .scenario-card,
.item-card, ...) working without rewrites. <html data-theme="dark">
preserves v2's dark visual identity; light-mode toggle is deferred.
DOM, JS logic, scenario data, and command pipelines are unchanged.
Also includes .gitleaks.toml at repo root (path allowlist for vendored
MANIFEST.json files — SHA-256 file hashes are not secrets) which was
missed in the previous commit due to global git ignore.
Docs updated:
- README.md (root): notes the vendoring sync script + ms-ai-architect
Playground subsection
- plugins/ms-ai-architect/README.md: new Playground section with sync
workflow and standalone guarantee
- plugins/ms-ai-architect/CLAUDE.md: Playground section updated with
vendored design-system details + new filename
Initial sync of shared/playground-design-system/ into
plugins/ms-ai-architect/playground/vendor/playground-design-system/
via scripts/sync-design-system.mjs.
Source commit: f1fecf39b8
Files: 25 (7 CSS + 11 fonts/licenses + 3 schemas + README + MANIFEST)
Vendored copy keeps the plugin standalone — playground will load CSS
from ./vendor/ regardless of where the plugin is installed.
Also adds .gitleaks.toml at repo root with a path allowlist for
vendored MANIFEST.json files (SHA-256 file hashes are not secrets).
Docs updated together with the playground HTML refactor that actually
consumes the vendored CSS (next commit). This commit is internal-only.
Phase 2 bulk replace produced a few factually wrong attributions where
real publicly known sector documents/datasets/personas were incorrectly
re-attributed to the fictional generic entity. Genericize those
references instead.
- ros-sector-checklists.md: V440 håndbok citation -> "sektorvise
faglige håndbøker"; tilsynsmyndighet list -> generic phrasing
- master-data-management-ai.md: NVDB row -> generic "sektor-/fagregistre"
- ai-center-of-excellence-setup.md: NVDB integration line -> generic
"sektorvise nasjonale registre"
- multimodal-prompt-engineering.md: system_message persona -> generic
"fagingeniør i norsk offentleg sektor"
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Tiny helper command for ad-hoc multi-session flows that don't run through
/ultraexecute-local. Writes .session-state.local.json so /ultracontinue
can resume in a fresh chat. Required args (next-brief-path, next-label) —
no inline prompt, headless-safe. Validates via session-state-validator
and prints the same 3-line narration that /ultracontinue Phase 3 uses
(SC-8 cross-project consistency).
Step 9 of /ultracontinue v3.3.0. README/CLAUDE updates land in Step 11.
Extends the PreCompact hook with a sibling block that refreshes
.session-state.local.json's updated_at when status is in_progress or
partial. Per-project: runs after the existing progress.json mutation,
inside the same loop iteration.
Design:
- Only refreshes existing state files; creation is the writer's job
(ultraexecute Phase 8 / 2.55 / 4 + future helper command).
- Monotonic guard: only updated_at is touched. project, status,
next_session_brief_path, next_session_label remain owned by the writer.
- Skips status in {completed, failed, stopped} — the latter two are
operator-action-required and silently bumping updated_at would mask
alert state.
- Always exit 0; never blocks compaction.
[skip-docs] rationale: README + CLAUDE.md updates land in Step 11.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three insertions in commands/ultraexecute-local.md so every session-end
path produces or refreshes .session-state.local.json (Handover 7):
- Phase 2.55 (Check 1, line ~376): write status=stopped on dirty-tree
pre-flight stop before parallel session-spawn
- Phase 4 (line ~773): write status=stopped when entry condition fails
- Phase 8 (line ~1151): canonical convergence — every completed/failed/
stopped/partial run refreshes the state file using atomicWriteJson +
validator verification
Phase 2.3 (validate exit) and Phase 5 (dry-run) intentionally skip the
write — neither path is resumable. Validator errors warn but never block
the run; progress.json remains authoritative.
[skip-docs] rationale: README + CLAUDE.md updates land in Step 11.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Reads .claude/projects/<project>/.session-state.local.json (Handover 7),
narrates a 3-line summary, and immediately begins executing the next
session — no interactive confirmation, headless-safe.
Phases:
- 0: --help (self-documenting per brief NFR)
- 1: resolve project dir (auto-discover via node -e enumeration)
- 2: validate via session-state-validator
- 3: narrate (project / next_session_label / brief path)
- 4: read brief and begin
- 5: stats
[skip-docs] rationale: README + CLAUDE.md updates land in Step 11 (Session
2b) per plan structure. Step 8 (docs:) updates HANDOVER-CONTRACTS.md and
the doc-consistency test pin in the same session.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Plain-language UX humanizer release. Default output of all 18 commands
now leads with prose; technical IDs surface at end-of-line as references
rather than headlines. Scanner internals are unchanged; humanization is
a pure output-time transform applied at the rendering layer.
Highlights:
- New scanner-lib modules: humanizer.mjs, humanizer-data.mjs (TRANSLATIONS
for 13 scanner prefixes)
- New --raw flag threaded through every CLI for byte-stable v5.0.0
verbatim output (--json unchanged from v5.0.0, also byte-stable)
- 5 user-impact categories, 5 action-language phrases, 3 relevance contexts
- Self-audit terminal output also humanized; --json path unchanged
- 21 command and agent templates updated for humanized rendering with
--raw passthrough
- 635 → 792 tests (+157) including SC-3 forbidden-words lint, SC-4
scenario read-test, SC-5/6/7 backwards-compat snapshots
Migration:
- Existing --json automation: zero changes required (envelope is
byte-stable with v5.0.0; humanizer fields are bypassed)
- stderr-scraping tooling: review default mode (now uses prose); pass
--raw for v5.0.0 verbatim
- No scanner-internal changes (IDs, severity ladders, scoring weights,
area scorecards all unchanged)
Verification:
- 792/792 tests pass
- self-audit configGrade A (97), pluginGrade A (100), readmeCheck passed
- README badge: tests-635+ → tests-792+
- Plugin README: add "What's New in v5.1.0" section with humanizer overview,
before/after example, plain-language vocabulary table, --raw flag docs.
Bump version badge 5.0.0 → 5.1.0. Add Version History row.
- Plugin CLAUDE.md: add humanizer.mjs + humanizer-data.mjs to Scanner Lib
table. Add "Plain-Language Output (v5.1.0)" section documenting output
modes, vocabularies, and Wave 5 lessons. Bump test count 635 → 792 across
52 test files.
- Marketplace root README: bump config-audit entry 5.0.0 → 5.1.0, update
one-line description to mention plain-language UX, add bullet for the
v5.1.0 humanizer, bump test count 635+ → 792+.
Test-normalizer hardening (consequence of growing CLAUDE.md):
walkClaudeMdCascade walks upward from the marketplace-medium fixture into
this plugin's own CLAUDE.md, so any docs edit ripples into
`scanners[*].activeConfig.claudeMdEstimatedTokens`. The v5.0.0 byte-stability
contract is about scanner internals being unchanged, not ancestor input
content being frozen. Normalizers in json-backcompat, raw-backcompat,
posture-humanizer, scan-orchestrator-humanizer, and snapshot-default-output
now strip claudeMdEstimatedTokens to <ANCESTOR_DERIVED>. The
default-output snapshot for scan-orchestrator was re-seeded via
UPDATE_SNAPSHOT=1 (intent: Wave 6 docs additions; humanizer prose
unchanged).
Verify:
- grep -E "5\.1\.0|v5\.1\.0" README.md CLAUDE.md ../../README.md | wc -l = 12
- node --test 'tests/**/*.test.mjs' = 792/792 pass
- self-audit configGrade A (97), pluginGrade A (100), readmeCheck.passed true
- Bump README test-count badge: 635 → 792 (matches filesystem after Wave 0–5)
- Wire formatSelfAudit() through humanizeEnvelope + humanizeFindings so the
terminal-output path renders humanized finding titles. The --json path is
unchanged — only the prose terminal render is humanized.
- readmeCheck.passed now returns true; configGrade A (97), pluginGrade A (100)
Three changes in one commit:
1. NEW lib/util/atomic-write.mjs — exports atomicWriteJson(path, obj),
the canonical tmp+rename pattern. Reused by pre-compact-flush.mjs and
(in subsequent steps) by the new session-state writer.
2. NEW tests/lib/atomic-write.test.mjs — 4 unit tests covering
round-trip, no-orphan-tmp, overwrite-atomic, pretty-print formatting.
3. REFACTOR hooks/scripts/pre-compact-flush.mjs — replace the inline
atomicWrite() with the imported atomicWriteJson(). Also fixes a
pre-existing syntax error (leading whitespace + stray --resume token
outside the comment block) that silently broke the hook from v3.1.0
onward — PreCompact runtime is fail-open and swallowed the error.
File reformatted with standard zero-indent JS.
163 → 167 tests, 0 fail.
Step 2 of /ultracontinue v3.3.0 (project 2026-05-01-ultracontinue).
Brief assumed *.local.* covered .session-state.local.json — only *.local.md
existed. Adding *.local.json before any state file can be created.
Step 1 of /ultracontinue v3.3.0 (project 2026-05-01-ultracontinue).
Wave 5 Step 16 — final wave step. Threads humanizer-aware rendering
rules through the three agent prompts that produce user-facing output,
and adds a shape test that locks the structure.
- agents/analyzer-agent.md: documents the humanizer envelope shape
(userImpactCategory, userActionLanguage, relevanceContext) in the
Input section; new "Humanizer-aware rendering rules" subsection
instructs the agent to: render humanized title/description/
recommendation verbatim, group findings by userImpactCategory, lead
each line with userActionLanguage, surface relevanceContext when
not affects-everyone, and skip jargon-translation subroutines.
--raw fallback documented (v5.0.0 verbatim severity prefiks).
- agents/planner-agent.md: documents the same vocabulary; instructs
the planner to consume humanized fields from the analysis report,
preserve titles verbatim, and order actions by both dependencies
AND userActionLanguage urgency. Translation duties explicitly
removed from the plan.
- agents/feature-gap-agent.md: replaces the inline t1/t2/t3/t4
tier-to-prose section ladder with userActionLanguage-driven
groupings ("Fix soon" → High Impact, "Fix when convenient" →
Worth Considering, "Optional cleanup"/"FYI" → Explore When Ready);
instructs skipping findings whose relevanceContext is
test-fixture-no-impact; --raw fallback documented.
tests/agents/agent-prompt-shape.test.mjs (new, +6 tests, 786 → 792):
- structural: humanized field reference + frontmatter preserved
- per-agent anchors: analyzer groups by userImpactCategory; planner
orders by userActionLanguage; feature-gap references
test-fixture-no-impact
- global: no "explain what {jargon} means" / "translate jargon" /
"jargon-translation duty" prose anywhere
Self-audit: Grade A unchanged (config 97/100, plugin 100/100).
