run-history-pipeline

name	run-history-pipeline
description	Mine this repository's git history for failure-mode practices and extract them as reusable skills, one step at a time, clearing context between steps (`/clear` in Claude Code, or clear context in Cursor) to prevent context blowup.

When to use this skill

Extract failure-mode practices from a repository's git history. Runs two steps in order, clearing context between them (/clear in Claude Code, or clear context in Cursor):

1. mine-history (/mine-history in Claude Code or @mine-history in Cursor) → reads git log, produces _analysis/directives.md, _analysis/architecture.md, and _analysis/incidents.md
2. synthesize (/synthesize in Claude Code or @synthesize in Cursor) → reads _analysis/incidents.md, produces _analysis/extracted_skills_history/*/SKILL.md + plugin.json

Supports --from [step-name] to start from a specific step, e.g. /run-history-pipeline --from synthesize (Claude Code) or @run-history-pipeline --from synthesize (Cursor).

Note: This pipeline is optional. If you only want to analyze the codebase's current design (not its history), run /run-analysis-pipeline (Claude Code) or @run-analysis-pipeline (Cursor) instead.

---

Step 1 — Read pipeline state

Check for _analysis/.history-pipeline-state. Read it if it exists.

State file format:

completed: mine-history
next: synthesize
remaining:
published: false

Parse completed, next, remaining, and published fields. If the file doesn't exist, all steps are pending and mine-history is next.

If --from [step-name] was passed:

• Set next to the specified step
• Set completed to all steps that precede it in the pipeline order
• Set remaining to all steps that follow it
• Proceed with this overridden state (do not require a state file to exist)

If the state file shows both steps completed and published: false (or published field absent):

"All 2 pipeline steps are done. Skills are in _analysis/extracted_skills_history/.

Run /curate-skills (Claude Code) or @curate-skills (Cursor) to review and clean up, then /publish-skills (Claude Code) or @publish-skills (Cursor) to create a PR."

Then stop.

If the state file shows all steps completed and published: true:

"Pipeline complete and skills already published."

"To re-run from a specific step: /run-history-pipeline --from [step-name] (Claude Code) or @run-history-pipeline --from [step-name] (Cursor)"

Then stop.

Tell the developer the current state:

Pipeline state:
  ✓ mine-history (done)
  → synthesize (next)

Use ✓ for completed, → for next (about to run), · for pending.

---

Step 2 — Run the current step

Read the SKILL.md for the current step from ~/.prism/skills/[step-name]/SKILL.md and execute it fully.

Step name mapping:

• mine-history → ~/.prism/skills/mine-history/SKILL.md
• synthesize → ~/.prism/skills/synthesize/SKILL.md

Important: Run the step as written — do not skip its human-in-the-loop stages, do not abbreviate its analysis, do not shortcut its confirmation checkpoints. The full interactive flow is the point.

If the developer provided context in Step 2, use it to inform how you approach the step — treat it as additional background that the step's skill didn't have.

---

Step 3 — Update state and prompt for /clear

After the step completes successfully:

1. Write the updated state to _analysis/.history-pipeline-state:

   • Move the just-completed step from next to completed
   • Set next to the first remaining step (if any)
   • Update remaining accordingly
   • Always include published: false if not already present

   Example after completing mine-history:

   completed: mine-history
   next: synthesize
   remaining:
   published: false
   

   After completing synthesize:

   completed: mine-history, synthesize
   next:
   remaining:
   published: false
   

2. If more steps remain:

   "---" "[step-name] complete." "" "Clear context (/clear in Claude Code, or clear context in Cursor), then invoke @run-history-pipeline (Cursor) or /run-history-pipeline (Claude Code) to continue with [next-step-name]."

3. If all 2 steps are done:

   Generate the report section:

   • Read _analysis/.meta for repository and source. Read every plugin.json under _analysis/extracted_skills_history/*/plugin.json.
   • For each file extract: name, category (join the array with , ), and description (trim at first TRIGGER occurrence)
   • Write or update _analysis/report.md:
     • If the file does not exist, create it
     • Replace content between <!-- history-start --> and <!-- history-end --> markers with the new section; if the markers are absent, append it
   • Section format:

     <!-- history-start -->
     # Extracted from: {repository} · {DD-MM-YYYY} · {source}
     
     ## Git History Pipeline
     
     | Skill | Category | Description |
     |-------|----------|-------------|
     | {name} | {category} | {description} |
     <!-- history-end -->
     

   • If _analysis/extracted_skills_history/ is empty or absent, write the table headers with a single row: | — | — | No skills extracted. |

   Then tell the developer:

   "---" "History pipeline complete." "" "N skill(s) written to _analysis/extracted_skills_history/. Report saved to _analysis/report.md." "" "Clear context (/clear in Claude Code, or clear context in Cursor), then invoke @curate-skills (Cursor) or /curate-skills (Claude Code) to review before publishing."