CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

babysit-pr

Use whenever an open PR needs shepherding to a merge — watching CI, weighing the agentic reviews, resolving merge conflicts, and merging once everything is green. Assumes the PR already exists (code is written, pushed, and `gh pr create` has run); this skill starts from there. Used by the `build-feature` skill's last step, and by any agent (e.g. a dedicated babysitter) whose job is watching a PR through to landing.

77

Quality

97%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Quality

Content

100%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is an exemplary instruction-only skill: executable commands in every phase, explicit validation checkpoints with feedback loops for each mergeStateStatus outcome, and no padding. Its self-contained structure needs no external reference files.

DimensionReasoningScore

Conciseness

The body is lean and assumes Claude's competence — no git/gh basics are explained, and repo-specific facts Claude cannot know ("an architecture review, a mutation review ... take minutes, not seconds") justify their tokens. Commands carry inline comments ("# run in the background; resolves when all checks finish") instead of surrounding prose, matching the 5 anchor; the one duplicated rebase snippet appears in two genuinely distinct contexts, so nothing is trimmable enough to drop it to 4.

5 / 5

Actionability

Fully executable throughout: "gh pr checks <pr> --watch --interval 30", "gh run watch <run-id> --interval 20 --exit-status", "gh pr view <pr> --json mergeStateStatus -q .mergeStateStatus", "npm run check && git push --force-with-lease", and "gh pr merge <pr> --squash --delete-branch". Every phase has copy-paste-ready commands and per-status branches cover the common cases, matching the 5 anchor rather than 4's 'minor gaps'.

5 / 5

Workflow Clarity

Three clearly sequenced phases (watch → close the loop → merge) with explicit validation ("check the PR is actually mergeable before you merge"), a feedback loop ("re-check `mergeStateStatus` until it's mergeable"), and error-recovery guidance ("never end the turn silently on a failed merge"). The destructive-operation cap does not apply because validation checkpoints are present, so it matches the 5 anchor, not 4.

5 / 5

Progressive Disclosure

No bundle files exist and none are needed: the skill is a single coherent workflow with numbered, well-organized sections where everything belongs inline and navigation is trivial. Per the simple-skill guidance, well-organized sections with no need for external references warrant a 5; there are no organization gaps that would justify 4.

5 / 5

Total

20

/

20

Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: concrete actions, an explicit 'Use whenever' trigger with a clear starting boundary, and third-person voice. The only weakness is trigger synonym coverage — 'pull request' spelled out and a few related natural phrases are absent.

DimensionReasoningScore

Specificity

The description lists four concrete actions — "watching CI, weighing the agentic reviews, resolving merge conflicts, and merging once everything is green" — which is comprehensive coverage of the PR-shepherding domain. It clearly matches the 5 anchor rather than 4, since no meaningful capability in the workflow is left out.

5 / 5

Completeness

It explicitly answers both questions: the "what" is the four concrete actions, and the "when" is "Use whenever an open PR needs shepherding to a merge", reinforced by a clear boundary ("Assumes the PR already exists ... this skill starts from there"). This matches the 5 anchor's requirement for concrete trigger phrases, not merely the implicit 'when' of anchor 4.

5 / 5

Trigger Term Quality

Good natural-language coverage: "open PR", "watching CI", "merge conflicts", "merging once everything is green", "watching a PR through to landing" are phrases users would actually say. It falls short of the 5 anchor because synonyms like "pull request" spelled out and terms like "checks" or "PR comments" are missing, but it is well above anchor 3's partial coverage.

4 / 5

Distinctiveness Conflict Risk

It carves a clear niche — post-creation PR shepherding — and explicitly delimits it ("Assumes the PR already exists (code is written, pushed, and `gh pr create` has run)"), minimizing conflict with code-writing, review, or PR-creation skills. The explicit boundary pushes it past anchor 4's 'minor overlap risk' to anchor 5.

5 / 5

Total

19

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

drufball/skylark

Path

plugins/skylark-builder/skills/babysit-pr/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.