CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

ce-babysit-pr

Babysits an open GitHub PR until merge-ready, first judging whether an old PR still belongs. Use when asked to watch a PR over time or update an old one — not for one-shot comment resolution or one CI failure. GitHub (incl. Enterprise) only.

Quality

89%

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

92%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.

An exceptionally disciplined orchestration skill body: lean invariant-first prose, a strictly ordered tick workflow with real validation gates and recovery loops, and clean one-level-deep progressive disclosure into a well-organized reference bundle. The only soft spot is that a handful of execution mechanics (atomic marks, snapshot invocation details) live entirely in reference files, leaving minor gaps if the body is read alone.

Suggestions

Inline one-line examples of the two most-used mechanics — the pr-snapshot invocation and the shared atomic mark syntax — so a reader can execute the core tick without opening references/tick.md first.

DimensionReasoningScore

Conciseness

The body is extremely lean: no concept explanations, no padding, and every clause encodes a real rule or boundary (e.g. 'never merges; offer stack-wide once if a confirmed managed stack needs work'). It fully assumes Claude's competence, matching anchor 5 ('every token earns its place') rather than 4, which requires trimmable over-explanation that is essentially absent.

5 / 5

Actionability

Concrete commands and skill invocations appear at decision points ('gh repo view', 'gh pr checkout <ref>', 'ce-resolve-pr-feedback mode:pipeline', 'gh stack merge') alongside exact state conditions ('manager_status == "confirmed"', 'open_needs_human == 0'). A few mechanics are deferred to reference files without inline specifics ('the shared atomic mark', the pr-snapshot invocation syntax), leaving minor gaps that fit anchor 4 ('mostly executable, minor gaps') rather than the fully copy-paste-ready anchor 5.

4 / 5

Workflow Clarity

A clearly sequenced four-step process with a strict 7-step ordering invariant per tick, explicit validation checkpoints (terminal check first, stale-SHA cancellation, checkout gate with 'no push access or dirty checkout → stop, say so'), exact stop-condition guards, and error-recovery loops (CI routed by cause with re-run/needs-human/ce-debug, detector re-arming). This matches anchor 5; the mutation-heavy workflow has its validation and feedback loops in place.

5 / 5

Progressive Disclosure

The body serves as an overview of invariants and routing, with twelve one-level-deep reference files each cited at the point of need and with a statement of what that file owns (e.g. branch-currency.md 'owns who updates a BEHIND branch'). All cited reference paths and the bundled pr-snapshot script exist in the bundle, and content is appropriately split — matching anchor 5.

5 / 5

Total

19

/

20

Passed

Description

82%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 with an explicit 'Use when' clause covering both positive and negative triggers, clear niche boundaries, and a platform scope. Its only weakness is that the 'what' stays at the headline level — the specific streams it reacts to (review comments, CI failures, branch currency) are left to the body rather than summarized as concrete capabilities.

DimensionReasoningScore

Specificity

Names the domain ('Babysits an open GitHub PR until merge-ready') plus two concrete actions ('judging whether an old PR still belongs', watching over time), but does not enumerate the underlying capabilities (handling review comments, CI failures, branch currency). This matches anchor 3 (domain and 1-2 concrete actions, not comprehensive) rather than 4, which requires several listed specific actions.

3 / 5

Completeness

Explicitly answers both questions: what ('Babysits an open GitHub PR until merge-ready, first judging whether an old PR still belongs') and when ('Use when asked to watch a PR over time or update an old one — not for one-shot comment resolution or one CI failure'), with concrete positive and negative trigger phrases. This is a clear anchor 5; anchor 4's weaker 'when' does not apply.

5 / 5

Trigger Term Quality

Includes natural phrases users would say — 'watch a PR over time', 'update an old one' — plus 'PR' and 'GitHub'. Missing common variations such as 'pull request' (only the abbreviation appears) and terms like 'CI' appear only in the negative scoping, so it fits anchor 4 (good coverage, a few natural terms missing) rather than 5.

4 / 5

Distinctiveness Conflict Risk

A clear niche (ongoing PR babysitting) with explicit exclusions ('not for one-shot comment resolution or one CI failure') that separate it from PR-feedback and CI-debug skills, plus a platform bound ('GitHub (incl. Enterprise) only'). Minimal conflict risk, matching anchor 5.

5 / 5

Total

17

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

EveryInc/compound-engineering-plugin

Path

skills/ce-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.