proof-checker

Rigorous mathematical proof verification and fixing workflow. Reads a LaTeX proof, identifies gaps via cross-model review (external reviewer backend, xhigh reasoning), fixes each gap with full derivations, re-reviews, and generates an audit report. Use when user says "检查证明", "verify proof", "proof check", "审证明", "check this proof", or wants rigorous mathematical verification of a theory paper.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

An exceptionally actionable, well-sequenced workflow with strong validation feedback loops, but it is held back by repeated restatement of the opt-in discipline across multiple sections and a monolithic structure with no bundled reference files to offload detail.

Suggestions

Consolidate the --deep-fix and --restatement-check opt-in guarantees (default-off, additive-only, no verdict crosstalk, 'unavailable' is non-blocking) into a single canonical section and reference it elsewhere instead of restating the rules four times.

Move the large PROOF_AUDIT.json schema blocks and the Phase 3.6 restatement-check algorithm into bundled reference files under references/ and link to them, reducing the inline bulk and giving the skill genuine one-level-deep progressive disclosure.

Provide the shared-references/*.md files (external-cadence, reviewer-routing, fan-out-pattern, acceptance-gate, assurance-contract, reviewer-independence) as part of the bundle or document them as external, so the signaled references resolve to real files.

Dimension	Reasoning	Score
Conciseness	The ~850-line body assumes Claude's competence (no basic-concept explanations) but restates the same opt-in/no-crosstalk discipline for --deep-fix and --restatement-check across four sections (Phase 1 addendum, Deep-Fix Mode, Submission Artifact Emission, Key Rules), so it could be tightened rather than being fully lean.	2 / 3
Actionability	Provides copy-paste-ready reviewer prompts, exact MCP tool-call conventions, concrete bash/pdflatex commands, full JSON schemas, and a fill-in fix-record template — fully executable guidance rather than abstract description.	3 / 3
Workflow Clarity	Phases 0 through 5.5 are explicitly sequenced with validation checkpoints (acceptance gate, per-fix compile check, re-review rounds, counterexample suite, regression audit) and clear validate→fix→retry feedback loops for a destructive/batch-style operation.	3 / 3
Progressive Disclosure	Sections and tables are well-organized and shared-references links are clearly signaled, but no bundle files exist (no references/scripts/assets or shared-references dir) and large schemas and the restatement-check algorithm live inline in a single monolithic file rather than being split into one-level-deep bundled references.	2 / 3
	Total	10 / 12 Passed

Description

100%

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 tight, third-person description that states concrete capabilities and gives explicit bilingual use-when triggers for a well-scoped niche. It earns full marks without padding.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Reads a LaTeX proof, identifies gaps via cross-model review', 'fixes each gap with full derivations, re-reviews, and generates an audit report' — matching the multi-action anchor rather than the vague single-domain anchor.	3 / 3
Completeness	Clearly answers both 'what' (verification + fixing workflow) and 'when' via an explicit 'Use when user says...' clause, satisfying the both-what-and-when anchor.	3 / 3
Trigger Term Quality	Natural user phrases are given explicitly in both English and Chinese ('verify proof', 'proof check', 'check this proof', '检查证明', '审证明'), covering common variations a user would actually say.	3 / 3
Distinctiveness Conflict Risk	The niche is narrow and explicit — 'rigorous mathematical verification of a theory paper' — with triggers unlikely to fire for unrelated skills, matching the clear-niche anchor.	3 / 3
	Total	12 / 12 Passed

Validation

75%

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 — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (855 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 4 suspicious	Warning

	Total	12 / 16 Passed

Repository: wanshuiyin/Auto-claude-code-research-in-sleep
Commit: 82076e5

Reviewed: 5 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.