重要前提
安装AI Skills的关键前提是:必须科学上网,且开启TUN模式,这一点至关重要,直接决定安装能否顺利完成,在此郑重提醒三遍:科学上网,科学上网,科学上网。查看完整安装教程 →
codebase-recon by outfitter-dev/agents
npx skills add https://github.com/outfitter-dev/agents --skill codebase-recon基于证据的调查 → 发现 → 置信度追踪的结论。
outfitter:patterns 技能outfitter:find-root-causes 技能outfitter:report-findings 技能<when_to_use>
不适用于:胡乱猜测、无证据的假设、调查前的结论
</when_to_use>
| 进度条 | 等级 | 名称 | 行动 |
|---|---|---|---|
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
░░░░░| 0 |
| 收集 |
| 收集初始证据 |
▓░░░░ | 1 | 概览 | 广泛扫描,发现表面模式 |
▓▓░░░ | 2 | 调查 | 深入探究,验证模式 |
▓▓▓░░ | 3 | 分析 | 交叉引用,填补空白 |
▓▓▓▓░ | 4 | 综合 | 关联发现,高置信度 |
▓▓▓▓▓ | 5 | 完成 | 交付发现 |
校准:0=0–19%,1=20–39%,2=40–59%,3=60–74%,4=75–89%,5=90–100%
诚实开始。清晰的代码库 + 聚焦的问题 → 等级 2–3。模糊或复杂 → 等级 0–1。
达到等级 4 时:"对发现结果有高置信度。再从一个角度验证即可达到完全确定。继续还是现在交付?"
低于等级 5 时:包含 △ 注意事项 部分。
证据优于假设 — 尽可能调查,仅在必须时才猜测。
多来源收集 — 代码、文档、测试、历史记录、网络研究、运行时行为。
多角度审视 — 从不同视角检查后再得出结论。
记录空白 — 用 △ 标记不确定性,追踪未知内容。
展示工作过程 — 发现结果包含支持证据,而不仅仅是结论。
校准置信度 — 区分事实、推论和假设。
<evidence_gathering>
先广后窄:
分层证据:
追踪线索:
</evidence_gathering>
<output_format>
每次证据收集步骤后输出:
{ 带支持证据的发现编号列表 }
{ 识别的重复主题或结构 }
{ 发现结果对当前问题的意义 }
总体:{进度条} {百分比}%
高置信度领域:
较低置信度领域:
假设:
空白:
未知:
</output_format>
<specialized_techniques>
加载技能以进行专门分析(参见步骤部分):
outfitter:patternsoutfitter:find-root-causesoutfitter:report-findings</specialized_techniques>
循环:收集 → 分析 → 更新置信度 → 下一步
每一步:
得出结论前(等级 4+):
检查证据质量:
检查完整性:
检查可交付成果:
始终:
绝不:
核心方法:
微技能(按需加载):
outfitter:patterns — 提取和验证模式outfitter:find-root-causes — 系统性问题诊断outfitter:report-findings — 多来源研究综合本地参考:
相关技能:
outfitter:pathfinding — 在分析前澄清需求outfitter:debugging — 结构化错误调查每周安装数
66
仓库
GitHub 星标
25
首次出现
Jan 26, 2026
安全审计
安装于
opencode62
cursor54
gemini-cli54
codex53
github-copilot51
claude-code49
Evidence-based investigation → findings → confidence-tracked conclusions.
outfitter:patterns skilloutfitter:find-root-causes skilloutfitter:report-findings skill<when_to_use>
NOT for: wild guessing, assumptions without evidence, conclusions before investigation
</when_to_use>
| Bar | Lvl | Name | Action |
|---|---|---|---|
░░░░░ | 0 | Gathering | Collect initial evidence |
▓░░░░ | 1 | Surveying | Broad scan, surface patterns |
▓▓░░░ | 2 | Investigating | Deep dive, verify patterns |
▓▓▓░░ | 3 | Analyzing | Cross-reference, fill gaps |
▓▓▓▓░ | 4 | Synthesizing | Connect findings, high confidence |
Calibration: 0=0–19%, 1=20–39%, 2=40–59%, 3=60–74%, 4=75–89%, 5=90–100%
Start honest. Clear codebase + focused question → level 2–3. Vague or complex → level 0–1.
At level 4: "High confidence in findings. One more angle would reach full certainty. Continue or deliver now?"
Below level 5: include △ Caveats section.
Evidence over assumption — investigate when you can, guess only when you must.
Multi-source gathering — code, docs, tests, history, web research, runtime behavior.
Multiple angles — examine from different perspectives before concluding.
Document gaps — flag uncertainty with △, track what's unknown.
Show your work — findings include supporting evidence, not just conclusions.
Calibrate confidence — distinguish fact from inference from assumption.
<evidence_gathering>
Start broad, then narrow:
Layer evidence:
Follow the trail:
</evidence_gathering>
<output_format>
After each evidence-gathering step emit:
{ numbered list of discoveries with supporting evidence }
{ recurring themes or structures identified }
{ what findings mean for the question at hand }
Overall: {BAR} {PERCENTAGE}%
High confidence areas:
Lower confidence areas:
Assumptions:
Gaps:
Unknowns:
</output_format>
<specialized_techniques>
Load skills for specialized analysis (see Steps section):
outfitter:patternsoutfitter:find-root-causesoutfitter:report-findings</specialized_techniques>
Loop: Gather → Analyze → Update Confidence → Next step
At each step:
Before concluding (level 4+):
Check evidence quality:
Check completeness:
Check deliverable:
ALWAYS:
NEVER:
Core methodology:
Micro-skills (load as needed):
outfitter:patterns — extracting and validating patternsoutfitter:find-root-causes — systematic problem diagnosisoutfitter:report-findings — multi-source research synthesisLocal references:
Related skills:
outfitter:pathfinding — clarifying requirements before analysisoutfitter:debugging — structured bug investigationWeekly Installs
66
Repository
GitHub Stars
25
First Seen
Jan 26, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
opencode62
cursor54
gemini-cli54
codex53
github-copilot51
claude-code49
Skills CLI 使用指南:AI Agent 技能包管理器安装与管理教程
48,700 周安装
▓▓▓▓▓| 5 |
| Concluded |
| Deliver findings |