CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

baoyu-wechat-summary

Summarizes WeChat group chat highlights into a structured digest using the local wx-cli binary (https://github.com/jackwener/wx-cli). Generates a normal digest by default; a roast (毒舌) version is opt-in. Maintains per-group history (history.json + history-digests.jsonl), per-user profiles, and per-group fact memory (memory.md) across runs, with privacy guardrails baked in. Use when the user asks to "总结群聊", "群聊精华", "群聊摘要", "summarize group chat", "group chat digest", mentions a WeChat group name with a time range, says "帮我看看 XX 群最近聊了什么", "XX 群有什么值得看的", or asks to "回溯画像" / "初始化画像" / "backfill profiles". Adds the roast version when the user says "毒舌版", "roast 版", "再来个毒舌的", or similar.

Quality

95%

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

SKILL.md
Quality
Evals
Security

Quality

Content

92%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 well-engineered, highly actionable workflow: concrete wx-cli commands, exact file formats, mandatory verification rounds with an audit checklist, and exemplary progressive disclosure through a when-to-read reference table with all bundle files present and one level deep. The only weakness is mild redundancy that inflates token cost without adding information.

Suggestions

State the output-formats.md load trigger once (keep the references table entry) instead of three times — the Step 3.9 preamble and the Round 2 'Load that file now' note both repeat it.

Compress the User Input Tools section to the priority rule, the numbered-message fallback, and the batching rule in one or two lines; the multi-paragraph elaboration restates what the rules already imply.

Merge the incremental-mode rules (currently described in Step 1 and again in Step 3) into a single location — Step 3's filtering step only — with Step 1 deferring to it.

DimensionReasoningScore

Conciseness

The body is dense with genuinely skill-specific material Claude could not know (wx-cli flags, history.json schema, filename sanitization, timestamp edge cases) and never explains known concepts, but a few passages could be trimmed: the User Input Tools section elaborates batching/priority rules at length, the instruction to read output-formats.md is repeated three times (references table, Step 3.9 preamble, Round 2 note), and incremental-mode rules appear in both Step 1 and Step 3. This fits anchor 4 (efficient, minor over-explanation) rather than anchor 5 (every token earns its place) due to that redundancy, and is above anchor 3 because none of it is generic padding.

4 / 5

Actionability

Guidance is fully executable with copy-paste-ready commands — `wx contacts --query "<group_name>" --json`, `wx history ... --since YYYY-MM-DD --until YYYY-MM-DD -n 5000 --json > "$TMPDIR/wx-messages.json"`, jq slicing examples, grep-based attribution audit — plus exact path templates, naming conventions, and concrete JSON/JSONL schemas for the history files. This matches the anchor-5 example of covering common cases completely, exceeding anchor 4 (which allows minor gaps).

5 / 5

Workflow Clarity

Steps 1→9 are clearly sequenced with the three-round generation loop, a mandatory Round 3 attribution audit (grep the raw message file, fix, record a verdict line), a completion checklist before declaring success, and explicit error-recovery paths (zero new messages → exit, any wx failure → setup.md, >7-day ranges → split into per-3-day digests). This matches anchor 5 (validation steps, feedback loops, checklist); the cosmetic numbering jump from Step 3.9 to Step 7 does not break the sequence, and the batch-operation validation cap does not apply because validation is present.

5 / 5

Progressive Disclosure

The body is a workflow skeleton with a references table giving each of six files a content summary and an explicit "何时读" (when-to-read) trigger, plus an upfront instruction to load files on demand rather than all at once. All six referenced files exist in references/, and a grep confirms none of them reference further files, so references are one level deep with clear navigation — a direct match to anchor 5.

5 / 5

Total

19

/

20

Passed

Description

100%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.

An exemplary description: concrete capabilities with named state files, both default and opt-in behavior stated, and an explicit 'Use when' clause carrying natural trigger phrases in Chinese and English. Third-person voice throughout, with no fluff or over-claims.

DimensionReasoningScore

Specificity

Lists multiple concrete actions with named artifacts — "Summarizes WeChat group chat highlights into a structured digest", "Generates a normal digest by default; a roast (毒舌) version is opt-in", "Maintains per-group history (history.json + history-digests.jsonl), per-user profiles, and per-group fact memory (memory.md)" — matching the comprehensive-coverage anchor. It is above anchor 4 because there are no material gaps: default behavior, the opt-in roast variant, state persistence, and backfill are all explicitly named.

5 / 5

Completeness

Explicitly answers both questions: "what" in the first two sentences (structured digest, normal vs roast versions, history/profiles/memory maintenance) and "when" via "Use when the user asks to ... or similar" with concrete quoted trigger phrases. This is a direct match to the anchor-5 example pattern, not merely present-but-vague (anchor 4).

5 / 5

Trigger Term Quality

Covers natural trigger phrases a user would actually say in both languages and registers — "总结群聊", "群聊精华", "群聊摘要", "summarize group chat", "group chat digest", "帮我看看 XX 群最近聊了什么", "XX 群有什么值得看的", "毒舌版", "再来个毒舌的", "回溯画像", "backfill profiles" — matching the comprehensive-synonyms anchor. A score of 4 would require missing natural terms, but colloquial variants, formal commands, and English equivalents are all present.

5 / 5

Distinctiveness Conflict Risk

The WeChat-group-chat niche is unambiguous and tied to a specific tool ("local wx-cli binary") and specific trigger contexts (group name with a time range, 回溯画像), giving it a clear niche with minimal overlap risk against other skills. Anchor 4 would require minor overlap with closely related skills, but no plausible conflicting skill shares these triggers.

5 / 5

Total

20

/

20

Passed

Validation

81%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

metadata_field

'metadata' should map string keys to string values

Warning

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

13

/

16

Passed

Repository

JimLiu/baoyu-skills

Path

skills/baoyu-wechat-summary/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.