← ClaudeAtlas

compass-summarize-tracelisted

After a compass-solve pipeline finishes, produces a compact written record for handoff, PR comment, or documentation — 7 sections (6 if Explore did not run) covering what was asked, assumed, weighed, run, produced, revised, and NOT done. Use when a finished multi-stage run needs to be captured: "summarize what we did", "write this up for the PR", "hand this off", "document the run", after compass-solve completes.
Anselmoo/werkstoff · ★ 1 · AI & Automation · score 67
Install: claude install-skill Anselmoo/werkstoff
# compass-summarize-trace Turn a completed `compass-solve` run into a fixed 7-section record, then **validate structure with the guard**. `GUARD="python3 ${CLAUDE_PLUGIN_ROOT}/scripts/compass.py"` ## Input Read `.compass/runs/<id>/state.json`, written by `compass-solve`'s state-write step — it is the source of `dag.stages`, `explore_ran`, and the content each section below summarizes. If it is missing, ask the user for the run id or the path to its state file; do not infer `dag.stages` or `explore_ran` from conversation context. ## The 7 sections (exact order) 1. **What was asked** 2. **What was assumed** 3. **Approaches weighed** — **omit this section ENTIRELY if Explore did not run.** Do not leave it empty; remove it. 4. **What ran** — a table listing **every stage from `dag.stages`**. 5. **What was produced** 6. **What was revised** — **always present, even when nothing was revised. If nothing changed, say so explicitly.** 7. **What was NOT done** Follow with a **completeness score table**: one row per section (7, or 6 if Explore was skipped), scoring whether it is present and non-empty against the rule for that section above. ``` | Section | Present | Notes | |----------------------|---------|-----------------------------------------| | What was asked | yes | | | What was assumed | yes | | | Approaches weighed | om