agentleFS
foryouself83foryouself83/harness-tier/skills/flow-uninstall/SKILL.md

flow-uninstall

Skill10 starsChanged 31 days agoApache-2.0View on GitHub

Remove harness-tier's wiring from this repo — the inverse of /flow-init. Deletes host-owned config, so it confirms first. Run BEFORE /plugin uninstall.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Remove what is no longer wanted by hand.

    line 60
  2. Distinctive:

    Commit what the run changed — it names each file and directory it touched.

    line 63
  3. Very specific:

    After cleanup, the user can /plugin uninstall harness-tier to remove the cached plugin.

    line 65

≈ “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

Flow-Uninstall — Remove host-side wiring

Undo what /flow-init installed in the host repo. /plugin uninstall harness-tier removes only the cache (the plugin outside the host); everything flow-init wrote into the host repo stays unless removed here.

Run this BEFORE /plugin uninstall. The cleanup runs the plugin's flow_init_setup.py --uninstall, reachable only through ${CLAUDE_PLUGIN_ROOT} — so uninstalling the plugin first strands the host files this skill exists to remove. Order is the whole point of the skill.

Interaction. Wherever this skill asks the user something, use the host's blocking question tool already in your tool list, matched by capability rather than by a host-specific name; if it is listed but not loaded, load it first with the host's tool-discovery primitive; only when no such tool is listed, or a question call errors, offer numbered options in chat and end your turn to wait for the reply — never answer the question yourself or skip it.

Execution

  1. Confirm (destructive) — deleting .claude/harness-tier/ removes host-owned files too: flow-config.yaml, webhooks (teams-webhooks.json is git-tracked/team-shared), the design-doc templates under templates/ with any edits made to them, and gate evidence. List what will be removed — saying the templates can be copied out first — and ask the user (structured choice) to confirm (default: no). Stop if declined.
  2. Run the cleanup (idempotent — match-then-skip, the inverse of /flow-init):
    python3 "${CLAUDE_PLUGIN_ROOT}/scripts/flow_init_setup.py" --uninstall
    Relay its report. It:
    • Unregisters the commit gate and the harness-tier marketplace from .claude/settings.json (preserves any other hooks), and deletes the file when nothing else is left in it.
    • Strips the harness-tier .gitignore lines — keeping .teams-webhooks.local.json, which may still guard a secret — and the CLAUDE.md harness-tier:teams managed block.
    • Unregisters the Codex gate from .codex/hooks.json and strips the managed block from AGENTS.md, whether or not codex is still listed under harnesses.
    • Deletes .claude/harness-tier/ (scripts, config, evidence, webhooks, templates) and .claude/rules/harness-tier/ (the plugin's rules an older /flow-init may have copied in — the host's own rules stay).
  3. Relay the manual follow-ups the script prints (it does not do these — they're destructive to user-owned files):
    • .pre-commit-config.yaml's teams-notify-push / static-analysis hooks are left in place (team customizations / comments). Remove by hand if desired.
    • The workflows it names under .github/workflows/, read from the files there:
      • the ones calling the deleted .claude/harness-tier/ without a guard fail — release.yml from every release tool, on each push to the prerelease branch;
      • wiki-verify.yml, doc-style.yml and srs-verify.yml guard their script and exit 0, verifying nothing while each still spends a runner on every push;
      • self-contained renders (branch-naming.yml, entropy-check.yml, unit-test.yml, api-contract.yml, e2e.yml, deploy*.yml) keep running and costing runner minutes. Remove what is no longer wanted by hand.
    • Disable the installed git hooks: pre-commit uninstall --hook-type pre-commit --hook-type commit-msg --hook-type pre-push.
    • Commit what the run changed — it names each file and directory it touched.
  4. After cleanup, the user can /plugin uninstall harness-tier to remove the cached plugin.

Critical rules

  1. Confirm before destroying — never delete .claude/harness-tier/ without explicit approval — ask the user (structured choice); it contains host-owned config/credentials/webhooks.
  2. Order matters — run before /plugin uninstall (the cleanup script lives in the plugin).
  3. Leave user-owned tool config alone — .pre-commit-config.yaml and installed git hooks are reported for manual removal, never auto-edited (avoids destroying team customizations).

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.