Chalk Labs
Need help deploying babysit?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/ozeron-babysit
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: devops
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 Nov 1, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Shell scripts
Network hosts it talks to 2
codeload.github.com· found in install commandgithub.com· found in code
Files it touches
~/.claude/skills/babysit
What's inside
- Skill name
babysit- Description
- Babysit a PR to merge-ready. Use when asked to babysit, ship, or land a PR, get its CI green, address or resolve review comments, or fix its merge conflicts.
- License
- MIT
- Instructions
- 60 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (8)
- SKILL.md
- bulk.md
- fetch-threads.sh · scripts/
- pr-state.sh · scripts/
- respond.sh · scripts/
- wait.sh · scripts/
- pr-state.test.sh · tests/
- wait.test.sh · tests/
SKILL.md (first lines)
# Babysit PR
Drive the PR to **merge-ready**: checks green, mergeable, zero unresolved threads. Each pass clears one blocker with the smallest fix, then waits.
`S=~/.claude/skills/babysit/scripts`. `$S/pr-state.sh <pr>` prints one JSON snapshot; `.next` names the blocker.
## Mode
`.mode` from the first snapshot decides what you may write:
- `own`: commit fixes, push, reply and resolve threads.
- `review` (someone else's PR): GitHub comments only. Threaded replies plus one `gh pr comment` with findings; threads stay open for the author.
History is append-only: commit on top and merge the base in.
## Loop
Snapshot, act on `.next`, repeat:
| `.next` | Action |
| --- | --- |
| `conflicts` | `git merge origin/<base>`; resolve keeping both sides' intent. Intents clash → `git merge --abort` and ask. |
| `ci` | Per `.checks.failing`: `gh run view <run> --log-failed`. Caused by this PR → smallest fix, narrowest test first. Otherwise merge the base in once; still red → report it as out of scope. CI config stays as is. |
| `threads` | [Threads](#threads). |
| `update` | Merge the base in. |
| `wait` | [Wait](#wait). |
| `ready` | Own draft, otherwise done: `gh pr ready <pr>` so review bots run, then wait. |
| `done` | [Exit](#exit). |
| `merged`, `closed` | Report and stop. |
A pass ends on a push followed by a wait, or on `done`.
## Threads
`$S/fetch-threads.sh <pr>` lists unresolved threads; read each body, path and line, nothing else. Bots can also leave findings in a review body ("outside diff range", nitpicks); read the latest bot review on the head. Check every bot finding against the code before acting on it.
Close each thread once its fix is on the remote (`git ls-remote origin <branch>` shows the fix SHA), or once you decline it with a reason:
`$S/respond.sh <pr> <rootCommentId> <threadId> resolve "<what changed> (<sha>)"`
Use `noresolve` and ask when it needs a product call, the code can't answer it, or the fix couldn't be verified. Review mode always uses `noresolve`. Over ~20 threads, offer [bulk mode](bulk.md).
## Scope guard
Before every push, snapshot and read `.scope` (it counts unpushed commits). `growth` over ~200 lines, or `newFiles` outside the PR's area → stop and ask. Once the user approves, `$S/pr-state.sh <pr> --rebaseline`.
## Wait
Start `$S/wait.sh <pr>` with the Monitor tool (`timeout_ms: 1800000`; re-arm on expiry). It waits for checks to register and settle, then prints one snapshot line. Every wait goes through it.
`done` also needs the review bots' pass on the latest head: `botsPending` empty, their new threads handled, and a review or summary from each bot. Comment-only bots (Codex) never appear in `botsPending`. A green bot check can hide a skipped or paused review; report a paused or rate-limited bot instead of waiting on it.
## Exit
- **Done**: one `gh pr comment` mapping each handled thread to its fixing SHA or decline reason, plus the head SHA, `threads: 0` and `.review` (a human approval is the one gate left to them). Then PushNotification: `PR #<n> merge-ready at <sha>`.
- **Blocked**: PushNotification with the one question, then stop.
## Friction
When this skill, a script, or a repo doc misleads you, append to `~/.agent-feedback/<repo>/<branch with / as _>.md` (the repo's AGENTS.md "Feedback scratchpad" format if present): `### <date> · <what> · blocked|slowed|wrong`, then `Hit:`, `Workaround:`, `Fix:`.From ozeron/skills, MIT.
More from ozeron/skills
- 0.524score
improve-codebase-architecture
ozeron
Surface architectural friction and propose deepening opportunities — refactors that turn shallow modules into deep ones, for testability and AI-navigability.
SkillBFreshness: BClaude Code +9 - 0.499score
codebase-design
ozeron
Shared vocabulary for designing deep modules.
SkillBFreshness: BClaude Code +9 - 0.488score
implement
ozeron
Implement a piece of work based on a spec or set of tickets.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 7, 2026 (27 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/ozeron/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from ozeron
- 0.524score
improve-codebase-architecture
ozeron
Surface architectural friction and propose deepening opportunities — refactors that turn shallow modules into deep ones, for testability and AI-navigability.
SkillBFreshness: BClaude Code +9 - 0.499score
codebase-design
ozeron
Shared vocabulary for designing deep modules.
SkillBFreshness: BClaude Code +9 - 0.488score
implement
ozeron
Implement a piece of work based on a spec or set of tickets.
SkillBFreshness: BClaude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get babysit running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
babysit skill. Agent Mart. https://agentmart.dev/skills/ozeron-babysit (last updated 2026-10-24)