Name: eval-cs-skill
Rating: 71.2 (1 reviews)
Author: codestable

eval-cs-skill

CodeStable skill 工程化闭环入口。触发：写/改一个 cs skill、评测 skill 效果、跨 model/agent 量化、优化 skill 提示词、把收敛结论固化回 skill。内部推进 author、eval、optimize、release。

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

85%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-architected orchestration skill body: clear staged workflow with strong validation discipline and clean progressive disclosure, with only minor conciseness and placeholder-substitution gaps in the densest sections.

Suggestions

Tighten the '项目内学习迁移实验' subsection by moving the most granular checkpoint/attestation edge-case rules into references/ and keeping only the gating invariants inline.

Resolve {skill_dir}/{skill} placeholders into a concrete resolution step or example so the tool-entry commands are copy-paste ready.

Consider a one-line per-stage 'when to enter this stage' trigger next to each state-machine row so the next-step logic is fully self-contained without consulting the protocol files.

Dimension	Reasoning	Score
Conciseness	Largely lean and assumes Claude's competence (no concept preamble, compact tables/diagrams), but the dense '项目内学习迁移实验' edge-case block has trimming opportunities, keeping it just below a 5.	4 / 5
Actionability	Provides concrete executable commands with flags (runner.py/optimize.py/regression.py/bump_version.py) and a state-machine table with file-path conditions, but uses {skill_dir}/{skill} placeholders and defers some flows to references/*/protocol.md, leaving minor gaps.	4 / 5
Workflow Clarity	Clear staged state machine with explicit validation checkpoints (provenance git-commit-before-LLM gate, [measured]/[soft]/[underpowered] tags, $50 cost guard with --confirm, keep/kill convergence, expand-sample-rerun feedback) and explicit exit conditions; the batch/experimental skill has validation, so the cap-at-3 does not apply.	5 / 5
Progressive Disclosure	SKILL.md is an overview that signals one-level-deep per-stage references (author/eval/optimize/release/autonomy protocol.md, optimize/support templates), explicitly instructs per-stage loading, and all referenced bundle files verified present.	5 / 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, specific description with explicit trigger guidance and a well-scoped niche; its only weakness is slightly technical trigger phrasing rather than the most natural user vocabulary.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete lifecycle actions ('写/改一个 cs skill', '评测 skill 效果', '跨 model/agent 量化', '优化 skill 提示词', '把收敛结论固化回 skill'), covering the full author→eval→optimize→release loop comprehensively.	5 / 5
Completeness	Explicitly answers both 'what' (the engineering loop and its internal stages) and 'when' via a concrete '触发' trigger clause, matching the anchor that requires both with concrete trigger phrases.	5 / 5
Trigger Term Quality	The '触发' clause gives good keyword coverage across stages, but phrasing leans technical/jargony ('收敛结论固化', '跨 model/agent 量化') and lacks more colloquial synonyms a user would naturally say.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear self-referential niche ('造 skill 的 skill', CodeStable skill lifecycle) with distinct triggers and minimal overlap risk against ordinary feature skills.	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
referenced_paths_exist	Referenced path issues: 10 deeper-than-1-level	Warning

	Total	14 / 16 Passed

Repository: codestable/CodeStable
Path: .claude/skills/eval-cs-skill/SKILL.md
Commit: 19c796d

Reviewed: 6 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.