evolving-config

Name: evolving-config
Author: alexei-led

6installs5forks

How to install evolving-config?

alexei-led/cc-thingz/evolving-config

$npx -y skills add alexei-led/cc-thingz --skill evolving-config

Installs into the current project.

Summary

Audit AI coding-agent configuration with local evidence first. Default to review-only. Apply fixes only when the user explicitly asks or passes --fix.

Files · 1

View on GitHub

SKILL.md

1	# Evolving Agent Configuration
2
3	Audit AI coding-agent configuration with local evidence first. Default to
4	review-only. Apply fixes only when the user explicitly asks or passes `--fix`.
5
6	## Claude tool use
7
8	- Use Read, Glob, and Grep for local inventory.
9	- Use WebFetch for official docs, changelogs, and exact source URLs.
10	- Use Perplexity only when official docs do not answer a current best-practice or ecosystem question.
11	- Use AskUserQuestion before ambiguous scope or risky fixes.
12	- Do not fetch broad web advice before reading local config.
13
14	## Read first
15
16	- `references/RUBRIC.md` for shared review dimensions and severity.
17	- `references/platforms/claude-code.md` for Claude Code surfaces.
18	- `references/platforms/codex.md` for Codex surfaces.
19	- `references/platforms/pi.md` for Pi surfaces.
20	- `references/apply-fixes.md` only in approved fix mode.
21
22	## Modes
23
24	Review-only is the default for prompts such as "review my config", "audit
25	config", "check setup", or "what should I improve".
26
27	Fix mode starts only when the user explicitly asks for changes or passes `--fix`.
28	Even then, ask before changing permissions, sandbox policy, hooks, MCP servers,
29	model routing, package installs, deletes, moves, broad rewrites, private config,
30	or managed settings.
31
32	## Workflow
33
34	1. Identify platform, config root, and mode.
35	2. If scope is ambiguous, list detected surfaces and ask which to audit.
36	3. Inventory relevant files with paths, sizes, and source/generated status.
37	4. Read current files before recommending changes.
38	5. Check the matching platform reference.
39	6. When the audit touches skills, agents, prompts, or package manifests, inspect thin-router risk, weak pointers to must-read support files, and whether plugin grouping forces unrelated instructions into startup context.
40	7. Fetch official docs only when syntax, feature availability, or deprecation status is uncertain.
41	8. Use broader web research only for source gaps or ecosystem comparisons.
42	9. Classify findings by impact and disruption.
43	10. In fix mode, apply only approved changes and verify.
44
45	## Scope
46
47	Review AI-agent config only:
48
49	- Claude Code settings, `CLAUDE.md`, skills, agents, commands, hooks, MCP, permissions.
50	- Codex config, `AGENTS.md`, profiles, sandbox, approvals, MCP, skills, subagents.
51	- Pi settings, packages, skills, extensions, prompts, themes, context files.
52	- Plugin/package manifests and source-to-generated export rules.
53
54	Do not review app runtime config, git hook hygiene, product docs, source-code
55	quality, or generated output as the source of truth.
56
57	## Priorities
58
59	Flag these first:
60
61	- unsafe permissions, sandbox bypasses, broad MCP access, or secret exposure
62	- unsupported or stale keys, tool names, hook events, or model names
63	- generated exports edited by hand instead of source files
64	- duplicate or overlapping skills, agents, hooks, prompts, or trigger descriptions
65	- bloated startup context or always-loaded instructions that should be on demand
66	- hooks or extensions that hide errors, block safe work, or run risky commands
67	- missing validation for config that produces skills, agents, hooks, or packages
68
69	## Output
70
71	```markdown
72	## Config Audit
73
74	Scope: <platforms/files>
75	Mode: review-only \| fix-approved
76	Sources: <local files and docs checked>
77	Confidence: high \| medium \| low
78
79	### Summary
80
81	- Files reviewed: N
82	- Generated files skipped: N
83	- Main risk: <one sentence>
84
85	### Critical
86
87	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
88
89	### Important
90
91	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
92
93	### Suggested
94
95	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
96
97	### Working Well
98
99	- <config that should stay as-is>
100
101	### Verification
102
103	- <command run or recommended>
104	```
105
106	Omit empty severity sections. If no findings are confirmed, say `No confirmed
107	findings.`
108
109	## Failure handling
110
111	- Ambiguous target: ask one scoped question before auditing.
112	- Missing official docs: use local evidence, lower confidence, and report the gap.
113	- Generated file target: report the source path and regeneration command instead of editing it.
114	- Secrets or private data: do not quote secret values; identify only the path and key name when needed.
115	- Validation failure after a fix: revert the change unless the user asks to keep it, then report the exact error.

People who liked this also viewed

6 picks

Type

lark-drive飞书云空间（云盘/云存储）：管理 Drive 文件和文件夹，包含上传/下载、创建文件夹、复制/移动/删除、查看元数据、评论/权限/订阅、标题、版本、飞书文档密级标签（secure labels）和本地文件导入。用户需要整理云盘目录、处理云空间资源 URL/token、判断链接类型/真实 token/标题，或导入 Word/Markdown/Excel/CSV/PPTX/.base 为 docx/sSkillsJul 2026382k16k
lark-sharedUse for lark-cli setup/auth tasks: auth login/status/logout, user vs bot identity, business-domain permissions (--domain, including all/docs/drive), missing scopes, revoking authorization, or handlingSkillsJul 2026382k16k
lark-whiteboard飞书画板：查询和编辑飞书云文档中的画板。支持导出画板为预览图片、导出原始节点结构、使用多种格式更新画板内容。当用户需要查看画板内容、导出画板图片、编辑画板时使用此 skill。不负责：飞书云文档内容编辑（lark-doc）、文档内嵌电子表格/Base（lark-sheets / lark-base）。SkillsJul 2026380k16k
lark-task飞书任务：管理任务、清单和任务智能体。创建待办任务、查看和更新任务状态、拆分子任务、组织任务清单、分配协作成员、上传任务附件、注册或注销任务智能体、更新任务智能体的主页数据、写入智能体任务记录。当用户需要创建待办事项、查看任务列表、跟踪任务进度、管理项目清单或给他人分配任务、为任务上传附件文件、注册注销任务智能体、更新智能体主页数据、写入任务记录时使用。SkillsJul 2026380k16k
lark-attendance飞书考勤打卡：查询自己的考勤打卡记录SkillsJul 2026330k16k
lark-apps妙搭（Spark/Miaoda）应用开发与托管：应用创建、本地全栈开发、云端生成迭代、创意设计（UI mockup / 可交互原型 / 线框图 / 落地页 / 仪表盘 / 幻灯片 deck / 视觉探索）、AI相关能力和飞书平台能力或者其他外部能力集成、日志/Trace/监控指标/PV/UV 查询、环境变量管理、应用角色与成员管理、自动化触发器（定时/记录变更/Webhook/飞书审批）。当用户要SkillsJul 2026227k16k

evolving-config

6installs5forks

View on GitHub

How to install evolving-config?

alexei-led/cc-thingz/evolving-config

$npx -y skills add alexei-led/cc-thingz --skill evolving-config

Installs into the current project.

Summary

Audit AI coding-agent configuration with local evidence first. Default to review-only. Apply fixes only when the user explicitly asks or passes --fix.

Files · 1

View on GitHub

SKILL.md

1	# Evolving Agent Configuration
2
3	Audit AI coding-agent configuration with local evidence first. Default to
4	review-only. Apply fixes only when the user explicitly asks or passes `--fix`.
5
6	## Claude tool use
7
8	- Use Read, Glob, and Grep for local inventory.
9	- Use WebFetch for official docs, changelogs, and exact source URLs.
10	- Use Perplexity only when official docs do not answer a current best-practice or ecosystem question.
11	- Use AskUserQuestion before ambiguous scope or risky fixes.
12	- Do not fetch broad web advice before reading local config.
13
14	## Read first
15
16	- `references/RUBRIC.md` for shared review dimensions and severity.
17	- `references/platforms/claude-code.md` for Claude Code surfaces.
18	- `references/platforms/codex.md` for Codex surfaces.
19	- `references/platforms/pi.md` for Pi surfaces.
20	- `references/apply-fixes.md` only in approved fix mode.
21
22	## Modes
23
24	Review-only is the default for prompts such as "review my config", "audit
25	config", "check setup", or "what should I improve".
26
27	Fix mode starts only when the user explicitly asks for changes or passes `--fix`.
28	Even then, ask before changing permissions, sandbox policy, hooks, MCP servers,
29	model routing, package installs, deletes, moves, broad rewrites, private config,
30	or managed settings.
31
32	## Workflow
33
34	1. Identify platform, config root, and mode.
35	2. If scope is ambiguous, list detected surfaces and ask which to audit.
36	3. Inventory relevant files with paths, sizes, and source/generated status.
37	4. Read current files before recommending changes.
38	5. Check the matching platform reference.
39	6. When the audit touches skills, agents, prompts, or package manifests, inspect thin-router risk, weak pointers to must-read support files, and whether plugin grouping forces unrelated instructions into startup context.
40	7. Fetch official docs only when syntax, feature availability, or deprecation status is uncertain.
41	8. Use broader web research only for source gaps or ecosystem comparisons.
42	9. Classify findings by impact and disruption.
43	10. In fix mode, apply only approved changes and verify.
44
45	## Scope
46
47	Review AI-agent config only:
48
49	- Claude Code settings, `CLAUDE.md`, skills, agents, commands, hooks, MCP, permissions.
50	- Codex config, `AGENTS.md`, profiles, sandbox, approvals, MCP, skills, subagents.
51	- Pi settings, packages, skills, extensions, prompts, themes, context files.
52	- Plugin/package manifests and source-to-generated export rules.
53
54	Do not review app runtime config, git hook hygiene, product docs, source-code
55	quality, or generated output as the source of truth.
56
57	## Priorities
58
59	Flag these first:
60
61	- unsafe permissions, sandbox bypasses, broad MCP access, or secret exposure
62	- unsupported or stale keys, tool names, hook events, or model names
63	- generated exports edited by hand instead of source files
64	- duplicate or overlapping skills, agents, hooks, prompts, or trigger descriptions
65	- bloated startup context or always-loaded instructions that should be on demand
66	- hooks or extensions that hide errors, block safe work, or run risky commands
67	- missing validation for config that produces skills, agents, hooks, or packages
68
69	## Output
70
71	```markdown
72	## Config Audit
73
74	Scope: <platforms/files>
75	Mode: review-only \| fix-approved
76	Sources: <local files and docs checked>
77	Confidence: high \| medium \| low
78
79	### Summary
80
81	- Files reviewed: N
82	- Generated files skipped: N
83	- Main risk: <one sentence>
84
85	### Critical
86
87	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
88
89	### Important
90
91	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
92
93	### Suggested
94
95	- `path:line` — issue. Evidence: <fact>. Fix: <action>.
96
97	### Working Well
98
99	- <config that should stay as-is>
100
101	### Verification
102
103	- <command run or recommended>
104	```
105
106	Omit empty severity sections. If no findings are confirmed, say `No confirmed
107	findings.`
108
109	## Failure handling
110
111	- Ambiguous target: ask one scoped question before auditing.
112	- Missing official docs: use local evidence, lower confidence, and report the gap.
113	- Generated file target: report the source path and regeneration command instead of editing it.
114	- Secrets or private data: do not quote secret values; identify only the path and key name when needed.
115	- Validation failure after a fix: revert the change unless the user asks to keep it, then report the exact error.

People who liked this also viewed

6 picks

Type