fact-check

Verifies claims in backlog items, skill documentation, or plugin content against primary sources using web lookups. Spawns parallel verification agents that must use WebFetch/WebSearch/gh — training data recall is explicitly rejected as evidence. Produces VERIFIED/REFUTED/INCONCLUSIVE verdicts with citations. Use when items are marked UNVERIFIED or when verifying tool API claims, CLI flags, or documented software behavior.

Quality

90%

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

88%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 highly actionable, well-structured process skill: exact prompt templates, verdict formats, validation loops (CoVe, lint-before-commit), and concrete post-action commands. Its weaknesses are minor — token cost of two mermaid diagrams that restate simple rules, and external reference links that cannot be verified from within the skill's directory.

Suggestions

Replace the two mermaid flowcharts with one-line bullets (e.g. '1–5 claims: single wave; 6+: sequential waves of 5') — the branching they encode is trivial and the diagrams cost significant tokens.

Add brief guidance for handling agent failures (agent dies, returns a malformed verdict) during wave execution so the collect-and-report step has an explicit recovery path.

Fix or justify the References section paths (e.g. '../../../plugins/development-harness/...') so they resolve relative to this skill, or replace them with short inline attributions.

Dimension	Reasoning	Score
Conciseness	The body is dominated by lean operational templates (evidence rules, agent prompt block, verdict format, report skeleton, exact post-action commands) with no explanations of concepts Claude already knows. The two mermaid flowcharts (~35 lines) encode trivial branching ("spawn waves of 5") that a one-line bullet would state more cheaply, which is a minor instance of over-explanation — the level 4 anchor rather than level 5.	4 / 5
Actionability	Fully executable guidance: the exact per-agent prompt template (CLAIM/SOURCE_FILE/PRIMARY_SOURCE/...), the exact verdict text block, a complete copy-paste-ready report format, and concrete commands ("uv run prek run --files .claude/backlog/", "git add .claude/backlog/ && git commit -m ..."). This is instruction-shaped and every step is directly runnable, matching the level 5 anchor.	5 / 5
Workflow Clarity	Clear sequence (parse input → extract/classify claims → wave-spawn agents → collect verdicts → report → post-actions) with explicit validation checkpoints: the CoVe step is a built-in verification/feedback loop, INCONCLUSIVE verdicts require stating the resolving next step, and lint runs before the commit. This matches the level 5 anchor of explicit validation steps and feedback loops for a batch operation.	5 / 5
Progressive Disclosure	Well-organized self-contained sections with clear headers, and all operational content is appropriately inline for a process skill. However, no bundle files (references/, scripts/, assets/) exist, and the four trailing reference links point to paths outside the skill directory (e.g. "../../../plugins/development-harness/skills/root-cause-tracing-process/SKILL.md") that do not resolve here — navigation to the cited sources is unclear, matching the level 4 'minor organization gaps' anchor rather than level 5.	4 / 5
	Total	18 / 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, named tools and outputs, third-person voice, and an explicit 'Use when' clause with distinctive triggers. Its only gap is missing a few natural synonyms users might say (e.g., fact-check, hallucination) that would make triggering even more reliable.

Suggestions

Add one or two natural user phrasings to the trigger clause, e.g. 'when the user asks to fact-check a claim or suspects hallucinated/fabricated API details', to broaden natural trigger coverage.

Consider mentioning the parallel-agent limit or report output in the description only if it changes when users would invoke the skill; otherwise leave as-is to avoid padding.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions with named mechanisms: "Verifies claims... against primary sources using web lookups", "Spawns parallel verification agents that must use WebFetch/WebSearch/gh", and "Produces VERIFIED/REFUTED/INCONCLUSIVE verdicts with citations". Coverage is comprehensive across the workflow (input types, method, output), matching the anchor for multiple specific concrete actions rather than the 'minor gaps' level 4 anchor.	5 / 5
Completeness	Explicitly answers both: what ("Verifies claims... using web lookups", "Spawns parallel verification agents", "Produces VERIFIED/REFUTED/INCONCLUSIVE verdicts with citations") and when ("Use when items are marked UNVERIFIED or when verifying tool API claims, CLI flags, or documented software behavior"). Both are concrete with explicit trigger phrases, matching the level 5 anchor exactly.	5 / 5
Trigger Term Quality	Good natural-term coverage: "UNVERIFIED", "verifying tool API claims, CLI flags, or documented software behavior". However, common user phrasings such as "fact-check", "is this claim true", or "hallucinated content" are absent, so it falls just short of the comprehensive synonym coverage of the level 5 anchor.	4 / 5
Distinctiveness Conflict Risk	A clear niche — primary-source claim verification — with distinctive trigger markers (UNVERIFIED backlog items) and named tools (WebFetch/WebSearch/gh) that no adjacent research or structural-check skill would claim. Conflict risk with generic research skills is minimal, matching the level 5 anchor.	5 / 5
	Total	19 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 4 suspicious	Warning

	Total	14 / 16 Passed

Repository: Jamie-BitFlight/claude_skills
Path: .claude/skills/fact-check/SKILL.md
Commit: 3e0b2af

Reviewed: 3 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.