| name | cursorqb-autopsy |
| description | Goal-backed Step 1.5 of the CursorQB planning workflow - an existing-project autopsy. Use after Step 1 (or directly) to inspect an existing/partially built repository and write Planner-docs/Autopsy.md, a 13-section technical feedback report that enriches Step 2 sub-plans. Runs automatically for existing/non-empty repositories and is skipped for empty or nearly empty ones. Launched automatically as a Cursor goal via the define-goal skill; only creates/updates Planner-docs/Autopsy.md. |
CursorQB Autopsy (Step 1.5, goal-backed)
Inspect an existing or partially built repository in detail and write a single autopsy report
that Step 2 uses as supporting feedback. This step never changes source code, never changes
Main-Planing.md, and only creates/updates Planner-docs/Autopsy.md. Read the full Step-1.5
spec in autopsy-planner.md (next to this skill) and follow it end to end.
Execution model
- In-session, goal-backed. Read the bundled
autopsy-planner.md spec and follow it. Register the goal via the Cursor define-goal skill (get_goal then create_goal); if the goal tool is unavailable, proceed under the in-context goal contract - behavior is identical.
- Reliability + validator. Resolve the plugin root by walking up to the folder containing
.cursor-plugin/plugin.json; read references/workflow-quality.md and follow it. The bundled validator is at <plugin-root>/scripts/validate_planner_docs.py.
- Sources of truth (read-only):
Planner-docs/Main-Planing.md (primary) plus repository evidence (README, manifests, CI, tests, docs, service/package folders, configs).
- Only writable file:
Planner-docs/Autopsy.md in the user's active workspace (cwd). Never touch source, Main-Planing.md, or any Step 2/3 file.
- Language: the autopsy report is written in English.
When to run vs skip
- Run for existing or partially built projects: the repository is not empty and contains meaningful evidence such as README, manifests, source/service/package directories, tests, docs, configs, or CI.
- Skip for new or nearly empty repositories: do not create
Planner-docs/Autopsy.md; report that Step 1.5 was skipped because there is not enough repository evidence, then stop. Step 2 continues without it.
Goal contract (declare before starting)
- Objective: For an existing project, produce
Planner-docs/Autopsy.md using the exact 14-heading structure (# Project Autopsy + 13 numbered sections) from autopsy-planner.md, grounded in repository evidence.
- Success evidence:
Autopsy.md exists with all 14 headings in order, in English, no secrets; git status shows only Planner-docs/Autopsy.md changed; OR Step 1.5 was correctly skipped for an empty repo (no Autopsy.md created).
- Scope bounds: only create/update
Planner-docs/Autopsy.md. No source/config/test edits. No secrets, tokens, or credentials. No commit, push, or PR.
- Stop condition: stop on Success, on a correct skip, or on a blocking read error.
Procedure
- Read
Main-Planing.md and inspect the repository with the read-only commands listed in autopsy-planner.md.
- If evidence is too thin (empty/new repo), skip and report; do not write
Autopsy.md.
- Otherwise build
Autopsy.md incrementally (create the file, then add sections) with all 14 headings, grounded in evidence; redact any secret-like value to <redacted>.
- Validate: run
python3 <plugin-root>/scripts/validate_planner_docs.py --root . --mode step2 --strict and confirm autopsy_exists=true with no Autopsy.md heading errors (fallback: manually confirm the 14 headings in order). Confirm only Planner-docs/Autopsy.md changed.
Output and handoff
Report: whether Step 1.5 succeeded, was skipped, or was blocked; whether Autopsy.md was created/updated; the highest-priority AUTOPSY-P0/P1 signals; how Step 2 should use the report; and confirmation that only Planner-docs/Autopsy.md changed. Control returns to cursorqb-planner (Gate 1 feedback covers both Main-Planing.md and Autopsy.md); Step 2 (cursorqb-subplanner) reads Autopsy.md as supporting feedback.
