open/config-audit
1
0
Fork 0
Code Issues Releases 1 Activity

fix(commands): stop assuming shell state survives between blocks

Browse source 
Dogfooding `plan` + `implement` against a throwaway config surfaced one root
defect with many arms: the command templates treat consecutive fenced blocks as
one shell. They are not. Every ```bash fence runs as its own Bash call in its own
process, so a variable set in one block is empty in the next, and `$$` is a
different PID (measured: 21710 vs 22109).

The planner agent confirmed the sharpest arm at runtime, reporting that
`Mode: $RAW_FLAG` "arrived literally unsubstituted" — `--raw` was documented in
three command files while being functionally dead. A machine sweep found the same
root in 20 places across 9 files, well past the two the written fasit predicted:

  - `$RAW_FLAG` read from non-shell agent prompts (analyze, plan, implement)
  - `$TMPFILE` read across blocks (tokens, manifest, whats-active,
    plugin-health) — each command could not read the file it had just written
  - `$GLOBAL_FLAG` across blocks (fix)
  - `$TODAY` never assigned in any block (campaign), passing
    `--reference-date ""` to a write CLI in six places
  - three `$$` temp paths handed to the Read tool (fix), which expands neither

All now follow the hardened drift.md pattern: a fixed literal path, or a
re-derivation inside each block that needs it.

Also fixed, all confirmed against ground truth rather than inferred:

  - `implement` printed a rollback ID it never captured (the timestamp lived only
    inside a command substitution) — the one message a user reads after a bad run
  - `plan` reported "No analysis results found" for valid sessions, because Read
    was pointed at a glob it cannot expand; now uses Glob and verifies the
    analysis report exists before spawning the agent
  - five phase commands wrote state.yaml with two of four required fields; since
    the agent writes all four, a follow-up write silently deleted the rest
  - `implement` promised rollback deletes created files; rollback deliberately
    leaves them (M-BUG-26 still open) — the doc, not the engine, was wrong
  - `implement` claimed a score delta with no pre-change measurement
  - `verifier-agent` was told to write a report it has no tool to write
  - dead `Task` tool name in always-loaded rule context; planner-agent template
    demonstrated the inline file content its own line 110 forbids

The sweeps land as tests/commands/command-shell-state-shape.test.mjs, verified
red before the fix and proven able to fail by reintroducing the defect. Two
existing tests asserted the old bash-block mechanism rather than the intent and
were updated. Suite 1449/0; frozen v5.0.0 snapshots and all scanner code
untouched.

Not fixed, deliberately: neither command scope-gates its actions to the audit
target. The generated plan included an edit to a real file under ~/.claude,
outside the throwaway target, because the skill/agent scanners are machine-wide.
That is a design change, not a side fix.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0195udHgCcFegzm7ecKku2Yc
This commit is contained in:
Kjell Tore Guttormsen 2026-08-01 20:12:17 +02:00
commit 09f817977c
17 changed files with 413 additions and 71 deletions
Download patch file Download diff file Expand all files Collapse all files

9
commands/interview.md
View file

@ -11,8 +11,8 @@ Gather user preferences to inform the action plan.
## IMPORTANT: Inline Execution Only
This command runs AskUserQuestion **directly in the main context** — NOT via a Task subagent.
AskUserQuestion requires synchronous terminal interaction and does not work when delegated to a Task subagent.
This command runs AskUserQuestion **directly in the main context** — NOT via an `Agent` subagent.
AskUserQuestion requires synchronous terminal interaction and does not work when delegated to an `Agent` subagent.
## Prerequisites
@ -44,7 +44,7 @@ AskUserQuestion requires synchronous terminal interaction and does not work when
`/config-audit discover` first" and exit. Read the session's `state.yaml` and
verify `completed_phases` contains `analyze`; if it doesn't, tell the user
analysis hasn't run yet and exit. Then read the analysis report for context.
2. **Conduct interview inline**: Use AskUserQuestion tool directly (NOT via Task). Adapt questions based on analysis findings.
2. **Conduct interview inline**: Use AskUserQuestion tool directly (never delegate it to a subagent via `Agent` — a subagent cannot hold the interactive turn). Adapt questions based on analysis findings.
3. **Save interview results**: Write to `~/.claude/config-audit/sessions/{session-id}/interview.md`
4. **Update state** (see state-management rule), with one bound specific to this
command: interview is optional and can be run against a session that already
@ -53,7 +53,8 @@ AskUserQuestion requires synchronous terminal interaction and does not work when
phase already in `completed_phases` — append `interview` only if it is absent,
leave `current_phase`/`next_phase` pointing at the furthest phase reached, and
tell the user the preferences will apply the next time `/config-audit plan`
runs. Rewinding a finished session is how its progress gets lost.
runs. Rewinding a finished session is how its progress gets lost. Always set
`updated_at` to the current timestamp, whichever branch above applies.
5. **Output summary**
## Interview Questions