try-fix

Attempts ONE alternative fix for a bug, tests it empirically, and reports results. ALWAYS explores a DIFFERENT approach from existing PR fixes. Use when CI or an agent needs to try independent fix alternatives. Invoke with problem description, test command, target files, and optional hints.

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

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 workflow — every step is executable with explicit checkpoints, deferred-throw restore guarantees, and drift detection. Its weaknesses are repetition of the same prohibitions across many steps and a monolithic ~650-line body that keeps detail in SKILL.md instead of the reference bundle.

Suggestions

Move the Step 6 self-review procedure and the Step 7.5 drift-detection details (including the PowerShell code blocks) into a references/self-review.md file, linking from the main workflow to cut SKILL.md roughly in half.

State the restore command and the git-command prohibition once in a dedicated 'Cleanup rules' section and reference it from Steps 2, 6, 7.5, 8, and 9 instead of repeating the full text each time.

Consolidate the repeated artifact-gate rationale (Steps 6, 7.5, 8, 9) into the Step 8 gate block, keeping the per-step mentions to one-line reminders.

Dimension	Reasoning	Score
Conciseness	The body is procedural rather than explanatory (no padding with concepts Claude already knows), but the restore command and the "NEVER use git checkout/restore/reset" prohibition are repeated in at least five places, and the artifact-gate rules are restated across Steps 2, 6, 7.5, 8, and 9 — noticeably more tightening opportunity than the 'minor instances' of the 4 anchor.	3 / 5
Actionability	Every step ships copy-paste-ready PowerShell: baseline establishment with Tee-Object, JSON validation with try/catch and count capture, drift detection that handles the $null empty-file edge case, and exact artifact filenames with content templates. Concrete examples (approach.md, findings JSON schema, result.txt) cover the common cases.	5 / 5
Workflow Clarity	An explicit 11-step sequence with validation checkpoints throughout (baseline verification via Select-String, diff-drift detection, an artifact existence gate whose throw is deferred so restore always runs) and feedback loops (3-iteration test loop, self-review refresh on drift, JSON re-validation) — matching the top anchor.	5 / 5
Progressive Disclosure	The two references (compile-errors.md, example-invocation.md) are real, one level deep, and clearly signaled at point of use, but the main file is ~650 lines with large inline blocks (the full Step 6 self-review procedure, Step 7.5 drift-detection code, example artifacts) that clearly belong in separate reference files — fitting the 'content that should be separate is inline' anchor.	3 / 5
	Total	16 / 20 Passed

Description

83%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: third-person, concrete about actions and inputs, with an explicit 'Use when' clause for its CI/agent orchestration niche. Trigger terms are slightly machinery-oriented, so a few natural user phrasings ("try another approach", "retry a fix") are missing.

Suggestions

Add one or two natural-language trigger synonyms (e.g., "when a first fix attempt failed and another independent approach should be tried") so conversational users match it as readily as CI orchestrators.

Mention one or two key outputs (e.g., self-review findings count, baseline restoration) in the description to round out capability coverage.

Dimension	Reasoning	Score
Specificity	"Attempts ONE alternative fix for a bug, tests it empirically, and reports results" plus "Invoke with problem description, test command, target files, and optional hints" names several concrete actions and concrete input types, but coverage has minor gaps (self-review, baseline restore, and artifact outputs are unstated). Fits the 'several specific actions; minor gaps' anchor rather than the comprehensive 5.	4 / 5
Completeness	Explicitly answers both what ("Attempts ONE alternative fix... tests it empirically, and reports results") and when ("Use when CI or an agent needs to try independent fix alternatives"), with concrete trigger phrasing and invocation inputs — matching the top anchor.	5 / 5
Trigger Term Quality	Terms like "alternative fix", "bug", "PR fixes", "test", and "CI" give good keyword coverage, but natural user phrasings such as "try another approach", "attempt", or "retry a fix" are missing, and triggers are framed for CI/agent orchestrators rather than conversational users.	4 / 5
Distinctiveness Conflict Risk	The alternative-fix-attempt niche is clearly distinct from code review, test-running, and summarization skills, but a generic "fix this bug" request could overlap with general debugging skills, leaving minor conflict risk.	4 / 5
	Total	17 / 20 Passed

Validation

93%

Warnings & errors only

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

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (653 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: dotnet/maui
Path: .github/skills/try-fix/SKILL.md
Commit: 7a5a5d6

Reviewed: 7 days ago

Table of Contents

Discovery Implementation Validation

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.