CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

autopilot-pr-triage-scheduler

Queue open microsoft/apm pull requests (community fixes, PRs without an issue, or PRs that close an issue) and fan them out through an isolated pool (default 2) of autopilot-pr-triage-worker sessions. Advisory only. Never merges, never assigns, never requests reviewers, never runs autopilot-pr-review-worker. Activate on a mixed PR list or queue-open when the job is PR triage. Works in a local session, Copilot App automation, Cloud Agent, Remote Agent, or Agentic Workflow.

Quality

86%

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

82%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.

A well-engineered, highly actionable scheduler skill: executable commands, explicit stop-conditions at every risky step, and proper externalization of logic into scripts and assets. The main trimmable costs are a rule repeated across four sections and the absence of any fix-and-retry recovery loop beyond fail-closed stops.

Suggestions

State the never-comment/assign/request-reviewers rule once (e.g., in Hard nos) and reference it from the Activation card and Invocation sections instead of repeating it four times.

Add a lightweight recovery step to the Procedure for slot failures (e.g., whether a failed worker slot is retried, skipped, or stops the run) so the batch drain has an explicit feedback loop rather than stop-only handling.

Link assets/label-contract.json where it is named in the Selection section and state in one line when the reader must consult it.

DimensionReasoningScore

Conciseness

Telegraphic, imperative, and free of any explanation of concepts Claude already knows ("Do not merge. Do not push.", "ASCII only."). The 4 rather than 5: the never-comment/assign/request-reviewers rule repeats roughly four times across sections, and the "Ownership writes" block mostly describes other skills' write paths rather than this scheduler's own operation.

4 / 5

Actionability

Fully executable guidance: copy-paste commands for both sweep and dispatch modes ("python <this-skill>/scripts/fetch_queue.py --kind pr --mode sweep --repo microsoft/apm > batch.json"), a concrete gh-unavailable fallback (--records-json / --labels-json), a fully specified queue-table schema with column definitions, and complete Enter/Exit card templates.

5 / 5

Workflow Clarity

A clear 4-step procedure with abundant fail-closed validation checkpoints — "Missing -> stop", "Missing table, missing column, or blank rationale -> stop. Do not spawn.", "A helper failure stops the run" — plus a mandatory pre-spawn queue table and an exit receipt. It sits below the 5 anchor because error handling is stop-only with no fix-and-retry recovery loop, and the final report does not verify spawned worker outcomes.

4 / 5

Progressive Disclosure

Good bundle structure: the 577-line fetch logic is correctly externalized to scripts/fetch_queue.py, fan-out details live in a clearly linked one-level-deep assets/fan-out-pool.md (verified to exist), and label rules are in assets/label-contract.json. Minor gaps keep it at 4: label-contract.json is named as canonical owner but never linked or explained when to read it, and the queue-table column spec is dense inline.

4 / 5

Total

17

/

20

Passed

Description

87%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 what, explicit when-triggers, and crisp boundary negations that make conflict with sibling scheduler/worker skills unlikely. The only softness is system-internal jargon ("queue-open", "agentic-workflow") mixed into otherwise natural trigger phrasing.

DimensionReasoningScore

Specificity

Names the domain (microsoft/apm PR triage) with several concrete actions — "Queue open microsoft/apm pull requests", "fan them out through an isolated pool (default 2)" — plus scope qualifiers ("community fixes, PRs without an issue, or PRs that close an issue") and explicit exclusions ("Never merges, never assigns, never requests reviewers"). Coverage of the full workflow (queue table, selection rules, fan-out mechanics) is thin, keeping it below the comprehensive 5 anchor.

4 / 5

Completeness

Clearly answers both: what — "Queue open microsoft/apm pull requests… and fan them out through an isolated pool… Advisory only" — and when, with an explicit trigger clause: "Activate on a mixed PR list or queue-open when the job is PR triage." Both halves are concrete, matching the 5 anchor rather than the 4 anchor where 'when' is only partially explicit.

5 / 5

Trigger Term Quality

Good natural keyword coverage — "PR triage", "pull requests", "community fixes", "PRs without an issue", "mixed PR list" — that a maintainer would plausibly say. Internal system jargon ("queue-open", "agentic-workflow", "Copilot App automation") crowds out a few natural synonyms, so it falls just short of the comprehensive 5 anchor.

4 / 5

Distinctiveness Conflict Risk

A clear niche — advisory PR triage for one repo — with explicit boundary negations ("never runs autopilot-pr-review-worker", "Never merges, never assigns") separating it from the review/merge workers. Minimal conflict risk with any other skill.

5 / 5

Total

18

/

20

Passed

Validation

87%

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

Validation — 14 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

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

Warning

relative_links

Relative link issues: 1 suspicious

Warning

Total

14

/

16

Passed

Repository

microsoft/apm

Path

.agents/skills/autopilot-pr-triage-scheduler/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.