done
Close an explicit repository task by shaping, landing, publishing, proving, and cleaning it up; also handles narrower commit, publish, and…
Skill by edmundmiller
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/edmundmiller-done
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: productivity
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/done
What's inside
- Skill name
done- Description
- Close an explicit repository task by shaping, landing, publishing, proving, and cleaning it up; also handles narrower commit, publish, and landing-verification requests without exceeding their authority.
- License
- MIT
- Instructions
- 77 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (12)
- SKILL.md
- cleanup.md · references/
- git.md · references/
- github-pr.md · references/
- jj.md · references/
- closeout.py · scripts/
- publish-clean.py · scripts/
- teardown-herdr-worktree.sh · scripts/
- test-verify-workspace-cleanup.sh · scripts/
- verify-jj-landing.sh · scripts/
- verify-landing.sh · scripts/
- verify-workspace-cleanup.sh · scripts/
SKILL.md (first lines)
# Done
Close the current repository task completely while preserving unrelated work.
Invoke for `/done`, bare `done` in an active task, or an explicit request to
land, publish, or verify task changes. Do not treat ordinary completion
language such as “I am done reviewing” as publication authority.
## Authority envelope
A bare `done` authorizes full closeout: shape, validate, integrate, publish,
prove, and clean up. A narrower request authorizes only its named stages. Honor
explicit PR, local-only, no-push, squash, merge, rebase, or fast-forward terms.
Never promote “commit and push,” “publish this branch,” or “verify landing” into
full cleanup.
Repository review rules and protected branches override the direct path.
Otherwise **Default to direct landing** on the actual default branch/bookmark.
Discover the destination and remote from live state; never assume `main` or
`origin` merely from convention.
## Completion invariant
Report `Done` only when:
1. Explicit task revisions are reviewable and repository-declared gates plus
focused behavioral checks pass.
2. Exact ancestry, per-revision patch equivalence, or aggregate task-range
patch equivalence proves the task landed on the actual default destination.
3. A writable authoritative remote equals the proved local default tip.
4. The recorded task workspace and owned branches are safely cleaned up.
`Done locally` is terminal only when no remote exists by design or the request
explicitly required local-only closeout. A clean feature workspace, successful
push, merged PR label, or passing build alone is never completion proof.
## Workflow
1. Record `active_directory=$(pwd -P)` before changing directories. Establish
the explicit task boundary from a receipt, recorded revisions, or a direct
user statement. Subjects and ancestry position are not task provenance.
2. Capture read-only evidence before mutations:
```bash
python3 "${HOME}/.agents/skills/done/scripts/closeout.py" snapshot \
--repo "$PWD" --active-directory "$active_directory" \
${receipt:+--receipt "$receipt"} >"$snapshot"
```
If no receipt exists, establish provenance and use `hey agent-adopt` before
the first closeout mutation.
3. Shape only task work. Leave unrelated dirt unstaged. Run repository gates
plus focused tests. Allow one unambiguous task-caused repair cycle; otherwise
record `Blocked`.
4. Run `jj root --ignore-working-copy`. Read [jj closeout](./references/jj.md)
when it succeeds; otherwise read [Git closeout](./references/git.md). Never
initialize jj during closeout.
5. If direct publication is forbidden or review is required, read the
[GitHub PR path](./references/github-pr.md). GitHub is the explicit hosted
provider; unsupported providers require authenticated discoverable tooling
or a precise blocker.
6. After landing proof, read [receipt, cleanup, and reporting](./references/cleanup.md).
## Finite outcomes
- `Done`: remote landing and cleanup proved; receipt terminal.
- `Done locally`: local-only-by-design landing and cleanup proved; receipt terminal.
- `Landed; cleanup deferred`: remote landing proved, but safe cleanup did not.
- `PR ready; merge pending`: required PR exists but gates remain after bounded waiting.
- `Local only`: publication was expected but unavailable or unauthorized.
- `Blocked`: scope, validation, conflict, identity, policy, or proof is unresolved.
Partial outcomes remain open. On the next invocation, re-read live local and
authoritative state, accept recorded evidence only when it still matches, and
resume at the first unmet stage. A false terminal claim records `false_done`.From edmundmiller/dotfiles, MIT.
More from edmundmiller/dotfiles
- 0.673score
build-audit-logs
edmundmiller
Build or review audit trails in TypeScript/JavaScript apps using evlog (pipelines, typed actions, denials, retention, compliance-style reviews).
SkillAFreshness: A★ 80Claude Code +9 - 0.671score
excalidraw-cli
edmundmiller
Create, inspect, update, and locally render Excalidraw scenes with the Excalidraw CLI.
SkillAFreshness: A★ 80Claude Code +9 - 0.671score
analyze-logs
HugoRCD
Analyze application logs from the .evlog/logs/ directory.
SkillAFreshness: A★ 80Claude Code +9 - 0.668score
code-search
edmundmiller
Compares semantic, text, and structural search strategies.
SkillAFreshness: A★ 80Claude Code +9 - 0.665score
verify
edmundmiller
Verify dotfiles changes at runtime without switching the system.
SkillAFreshness: A★ 80Claude Code +9 - 0.636score
challenging-expert-assumptions
edmundmiller
Examines the assumptions and evidence behind categorical technical claims.
SkillBFreshness: B★ 80Claude Code +9 - 0.636score
auditing-decision-data
edmundmiller
Checks whether observations support a consequential decision.
SkillBFreshness: B★ 80Claude Code +9 - 0.633score
derive-web-api-client
edmundmiller
Derives a task-focused API client from authorized HAR/browser traffic.
SkillBFreshness: B★ 80Claude Code +9 - 0.633score
hass-config-flow
edmundmiller
Uses Home Assistant APIs on NixOS for integration config flows, entity/device inspection, and token setup.
SkillBFreshness: B★ 80Claude Code +9 - 0.633score
honcho-integration
edmundmiller
Integrates Honcho memory into Python or TypeScript agents.
SkillBFreshness: B★ 80Claude Code +9 - 0.633score
home-assistant-best-practices
edmundmiller
Selects Home Assistant automation, helper, device-control, and dashboard patterns.
SkillBFreshness: B★ 80Claude Code +9 - 0.633score
hass-declarative
edmundmiller
Manages Nix-owned HA automations, scenes, scripts, and entity identity.
SkillBFreshness: B★ 80Claude Code +9
Signals
- Stars
- 80
- Forks
- 6
- Open issues
- 4
- Last commit (this folder)
- Aug 31, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/edmundmiller/dotfiles
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from edmundmiller
- 0.673score
build-audit-logs
edmundmiller
Build or review audit trails in TypeScript/JavaScript apps using evlog (pipelines, typed actions, denials, retention, compliance-style reviews).
SkillAFreshness: A★ 80Claude Code +9 - 0.671score
excalidraw-cli
edmundmiller
Create, inspect, update, and locally render Excalidraw scenes with the Excalidraw CLI.
SkillAFreshness: A★ 80Claude Code +9 - 0.668score
code-search
edmundmiller
Compares semantic, text, and structural search strategies.
SkillAFreshness: A★ 80Claude Code +9 - 0.665score
verify
edmundmiller
Verify dotfiles changes at runtime without switching the system.
SkillAFreshness: A★ 80Claude Code +9 - 0.636score
challenging-expert-assumptions
edmundmiller
Examines the assumptions and evidence behind categorical technical claims.
SkillBFreshness: B★ 80Claude Code +9 - 0.636score
auditing-decision-data
edmundmiller
Checks whether observations support a consequential decision.
SkillBFreshness: B★ 80Claude Code +9
Cite this page
done skill. Agent Mart. https://agentmart.dev/skills/edmundmiller-done (last updated 2026-10-24)