agentleFS
p10ns11yp10ns11y/skills/cursor-transcript-harvest/SKILL.md

cursor-transcript-harvest

Skill2 starsChanged 40 days agoMITView on GitHub

Harvest Cursor, Grok Build, Kilo, and Cline agent transcripts into agent-prompt-tuning-lab: pnpm harvest:all, seed-manifest, env vars, and gold tagging. Use when collecting or re-indexing local agent JSONL for this repo.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run on the host where agent data lives (~/.cursor/projects, ~/.grok/sessions, VS Code/Cursor globalStorage), not inside a devcontainer-only …

    line 11
  2. Very specific:

    Prefer distill for workflow/skill/rule drafts (docs/DISTILL.md).

    line 63
  3. Very specific:

    Do not commit data/raw, data/processed, data/distill, zips, or data/manifest.jsonl.

    line 67

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Agent transcript harvest

Run on the host where agent data lives (~/.cursor/projects, ~/.grok/sessions, VS Code/Cursor globalStorage), not inside a devcontainer-only shell unless harvesting those paths.

Full harvest → distill

cd /path/to/agent-prompt-tuning-lab   # repo root
pnpm harvest:all
pnpm normalize          # --source all
pnpm split

pnpm rhai-host:build
pnpm distill-sessions -- --pilot --llm grok
pnpm distill-workflows -- --run-dir data/distill/<run-id>
# review drafts/, then:
pnpm install-distill -- --run-dir data/distill/<run-id> --profile personal-skills

harvest:all = Cursor host --all --unpack + devcontainer unpack + Grok/Kilo/Cline convert+unpack.

Cursor subagent files land at:

data/raw/host/<YYYYMMDD>/<workspace-slug>/<parent-session-id>/subagents/<subagent-id>.jsonl

Grok/Kilo/Cline land under data/raw/{grok,kilo,cline}/… as Cursor-shaped JSONL. See docs/AGENT_SOURCES.md.

Selective harvest

pnpm harvest:cursor
pnpm harvest:agents
pnpm harvest:grok
pnpm harvest:kilo
pnpm harvest:cline
pnpm harvest:host -- --all --unpack
CURSOR_REPO_NAMES=my-app,other-repo pnpm harvest:host -- --unpack
pnpm harvest:devcontainer -- --unpack
VariablePurpose
CURSOR_PROJECTS_HOSTOverride ~/.cursor/projects
CURSOR_REPO_NAMESComma-separated slug fragments
CURSOR_TRANSCRIPTS_DEVCONTAINEROverride devcontainer transcripts path
GROK_SESSIONS_ROOTOverride ~/.grok/sessions
KILO_TASKS_ROOTOverride Kilo tasks/ directory
CLINE_TASKS_ROOTOverride Cline/Roo tasks/ directory

After harvest

  1. pnpm seed-manifest — usually auto-run after unpack; appends manifest rows (parent_session_id for subagents; agent from sidecars).
  2. pnpm normalize — default --source all. Use pnpm normalize:host for Cursor-only.
  3. Optional gold tags: pnpm tag-manifest -- --tag gold --session-id <uuid>.
  4. Prefer distill for workflow/skill/rule drafts (docs/DISTILL.md).

Privacy

Do not commit data/raw, data/processed, data/distill, zips, or data/manifest.jsonl. See .cursor/rules/data-privacy.mdc.

Legacy: insights / suggest-artifacts

Secondary path (histograms + bundle drafts): pnpm insights, then node scripts/suggest-artifacts.mjs --bundle personal --llm grok, review, --apply, then node scripts/install-artifacts.mjs --target <repo> --bundle personal. Skip niche ideas — see docs/DEFERRED_SKILLS.md.

Reference

  • docs/PIPELINE.md
  • docs/DISTILL.md
  • docs/AGENT_SOURCES.md
  • docs/GOLD_SESSIONS.md
  • docs/DEFERRED_SKILLS.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.