CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

autofix-pr

Watch a pull request and fix it from a worktree of its own (or, with --in-place, the user's own checkout) - CI failures, review feedback from trusted authors and merge conflicts are fixed, pushed to the PR branch and answered on the PR under the user's git and gh identity. Usage - /autofix-pr [<pr>|<pr-url>] [--once] [--in-place] [--every <minutes>] [--no-reply] [--yes] [instructions], /autofix-pr status [<pr>], /autofix-pr resume [<pr>], /autofix-pr stop [<pr>].

Quality

80%

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

Fix and improve this skill with Tessl

tessl review fix ./packages/core/src/skills/bundled/autofix-pr/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

77%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 actionable, rigorously validated state-machine procedure where every failure path has a specified next step — the workflow and actionability are near-ideal for a complex orchestration skill. Its weaknesses are density (run-on conditional paragraphs that resist quick parsing) and a total absence of progressive disclosure, with all reference-worthy material inlined in one monolithic file.

Suggestions

Move the refusal-code catalog (## Refusals), the fix-agent brief template, and the In-place variant into reference files under references/, keeping SKILL.md as an overview that points to them one level deep.

Break the giant single-paragraph steps (notably Start steps 3 and 5) into shorter bullets or a decision table pairing each condition (localUnpushedCommits, checkedRecently, wasEnded, etc.) with its required action.

Consolidate the repeated gate-matching incantation ("a task whose gate contains autofix-pr snapshot <n> --gate") into one named helper description defined once and referenced by short label elsewhere.

DimensionReasoningScore

Conciseness

There is no concept-teaching padding (nothing explains what a PR or CI is) and nearly every sentence carries operational detail, so it is above 2; but the ~330-line body packs conditions into enormous run-on paragraphs (Start steps 3 and 5 are ~400-word single paragraphs) that could be markedly tightened into tables, bullets, or shorter sentences — matching the 3 anchor (mostly efficient but could be tightened) rather than 4, where only minor trimming would be needed.

3 / 5

Actionability

Guidance is fully executable: exact commands ("${QWEN_CODE_CLI:-qwen}" autofix-pr prepare <n> --repo <owner>/<repo> --host <host>), explicit parameter mappings (--take-over is take_over: true), a complete fill-in brief template, and a concrete JSON reply-plan example. This matches the 5 anchor: copy-paste ready and covering the common cases.

5 / 5

Workflow Clarity

Start, Round, Stop, Status, Resume, and After-a-background-fix are each numbered and ordered, with explicit validation ("Commit only when they pass", verified: false discards the worktree) and pervasive feedback loops (retry limits for busy/partial, resync restarts once per turn, fix_plan correct-and-rerun, refusal next codes routing every failure to a specified action). Validation is present for destructive operations, so no cap applies; this matches the 5 anchor (explicit validation steps, feedback loops for error recovery).

5 / 5

Progressive Disclosure

No bundle files exist and all content is inline in a single 330-line file. Real section structure is present (## Refusals, ## Arguments, ## In-place, ## Trust, etc.) so it is above the 2 anchor's "minimal structure", but content that clearly belongs in separate files — the refusal-code catalog, the fix-agent brief template, the in-place variant, the status-interpretation rules — is inlined, matching the 3 anchor.

3 / 5

Total

16

/

20

Passed

Description

75%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 highly specific, distinctive description that concretely states what the skill does in third person with exact subcommand usage. Its one structural weakness is the absence of any explicit 'Use when...' trigger guidance, which both caps completeness and leaves a few natural trigger phrases unsaid.

Suggestions

Add an explicit trigger clause, e.g., "Use when the user asks to fix a PR's failing checks, address review comments, or resolve merge conflicts automatically".

Include a few natural synonym phrases users would say, such as "failing checks", "broken build", or "review comments", alongside the existing terminology.

DimensionReasoningScore

Specificity

The description lists multiple specific concrete actions — "Watch a pull request and fix it", "CI failures, review feedback from trusted authors and merge conflicts are fixed, pushed to the PR branch and answered on the PR under the user's git and gh identity" — comprehensively covering the skill's capability surface including the worktree/in-place modes and all subcommands. It matches the 5 anchor (comprehensive coverage) rather than 4, which expects minor gaps.

5 / 5

Completeness

The "what" is clearly and concretely answered, but there is no "Use when..." clause or equivalent explicit trigger guidance — only a "Usage -" line of invocation syntax, which is not trigger guidance. Per the rubric guideline, a missing 'Use when' clause caps completeness at 3 (clear 'what', 'when' missing or only weakly implied).

3 / 5

Trigger Term Quality

Natural keywords users would say are present: "pull request", "PR", "CI failures", "review feedback", "merge conflicts", "fix", "pushed". A few natural variants are missing (e.g., "failing checks", "broken build", "review comments"), so it sits at the 4 anchor (good coverage, a few natural terms missing) rather than 5's comprehensive synonym coverage, and well above 3 since several strong natural phrases appear.

4 / 5

Distinctiveness Conflict Risk

It occupies a clear niche — autonomously watching and fixing a pull request via a dedicated autofix-pr helper — with distinct triggers (PR watching, CI failures, trusted review feedback) that no generic skill would claim, matching the 5 anchor (clear niche, minimal conflict risk).

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

QwenLM/qwen-code

Path

packages/core/src/skills/bundled/autofix-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.