Tessl • The management layer for AI agent skills

SkillPublic

convention-graph-discovery

约定图发现方法论：进一个 repo 先识别 repo-specific conventions，再定义 domain/extractor 接 Convention Graph 引擎。Use when: 进入陌生 repo、要画约定图、要找“改 X 影响谁”的约定层关联、 F242/Convention Graph Layer 工作。Not for: 普通符号跳转/LSP、文档索引检索、记忆图谱、直接使用 codegraph/GitNexus。Output: domain 定义 + extractor 计划 + gap/freshness/provenance 报告。 GOTCHA: 沉淀的是“怎么画图”的方法，不是把 cat-cafe 的 extractor 硬搬到所有 repo。

Quality

89%

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 disciplined, token-efficient single-file skill with an explicit sequenced protocol, executable CLI commands, and strong validation/feedback loops (freshness, negative fixtures, baseline comparison). The main gaps are the absence of a concrete extractor/domain-plugin code example and the lack of any reference-file split for the tabular detail sections.

Suggestions

Add a minimal worked example of a domain plugin definition and extractor skeleton so step 3-4 of the protocol become executable rather than artifact-checklist only.

Move the Quick Reference and Common Mistakes tables into a one-level-deep reference file (e.g., references/protocol-checklist.md) to keep SKILL.md as a lean overview.

Gloss project-internal terms (F242, cat-cafe, 猫) on first use so the skill is self-contained for readers entering a fresh repo.

Dimension	Reasoning	Score
Conciseness	The body is lean and dense — compact tables for Common Mistakes (错误/后果/修复) and Quick Reference, a 7-step protocol with zero padding, and it assumes competence ('LSP 看符号，grep 看文本；约定图看"这个 repo 的规矩"') without explaining concepts Claude already knows.	5 / 5
Actionability	The 'Product Entry / Commands' section gives copy-paste-ready pnpm CLI invocations with placeholder substitution, expected JSON output fields, and stale-freshness handling, but the core '定义 domain / 写 extractor' steps specify required artifacts (domainId, node kinds, negative fixtures) without a concrete domain-plugin or extractor code example — minor gaps keep it below fully executable.	4 / 5
Workflow Clarity	The 7-step Discovery Protocol is clearly sequenced with explicit validation steps (baseline comparison against grep/LSP, mandatory gap reporting) and feedback loops ('如果 freshness.stale=true… 先重跑 convention-graph:index'), reinforced by the Pressure Test checklist (negative fixtures, stale-on-change, dogfood diff).	5 / 5
Progressive Disclosure	No bundle files exist and the ~98-line body is well-organized with clear sections, so all content is appropriately inline; however, it exceeds the simple-skill threshold and reference-worthy material (Quick Reference, Common Mistakes tables, domain definitions) is inlined rather than split into one-level-deep reference files.	4 / 5
	Total	18 / 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, well-scoped description that explicitly states what the skill does, when to use it, when not to use it, and what it outputs, with concrete trigger phrases in both Chinese and English. Its only weaknesses are reliance on project-internal jargon (F242, cat-cafe) and methodology-level phrasing where a more concrete action list would help.

Suggestions

Replace or gloss internal jargon (F242, cat-cafe, 猫) with universally understandable terms so the description works for readers outside the originating project.

Add one or two English keyword variations (e.g., 'convention layer', 'impact analysis', 'who consumes X') to broaden natural trigger coverage.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions and named outputs ('进一个 repo 先识别 repo-specific conventions，再定义 domain/extractor', 'Output: domain 定义 + extractor 计划 + gap/freshness/provenance 报告'), though the methodology-level framing ('接 Convention Graph 引擎') leaves minor coverage gaps versus a fully comprehensive action list.	4 / 5
Completeness	Explicitly answers both 'what' (identify conventions, define domain/extractor, output gap/freshness/provenance report) and 'when' ('Use when: 进入陌生 repo、要画约定图…') with concrete trigger phrases, and adds explicit negative scope ('Not for: …').	5 / 5
Trigger Term Quality	Includes multiple natural trigger phrases users would actually say ('进入陌生 repo', '要画约定图', '改 X 影响谁', '顺藤摸瓜', 'convention graph'), but leans on internal jargon (F242, cat-cafe) and misses common English variations, so coverage is good rather than comprehensive.	4 / 5
Distinctiveness Conflict Risk	A clear niche (convention-layer graph with provenance/freshness) is explicitly fenced off from every neighboring capability ('Not for: 普通符号跳转/LSP、文档索引检索、记忆图谱、直接使用 codegraph/GitNexus'), minimizing wrong-skill trigger risk.	5 / 5
	Total	18 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: zts212653/clowder-ai
Path: cat-cafe-skills/convention-graph-discovery/SKILL.md
Revision: 14e32aa

Reviewed: 28 days ago

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.

convention-graph-discovery

Table of Contents

Is this your skill?