Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/ronronner02/codepilot-agent/spec-rule-minernpx skills add ronronner02/codepilot-agent --skill spec-rule-minergit clone --depth 1 https://github.com/ronronner02/codepilot-agentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ronronner02/codepilot-agent/spec-rule-miner)<a href="https://agentmods.dev/skills/ronronner02/codepilot-agent/spec-rule-miner"><img src="https://agentmods.dev/badge/skills/ronronner02/codepilot-agent/spec-rule-miner.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00118 | $0.02020 |
| Opus 5 | $0.00059 | $0.01010 |
| Sonnet 5 | $0.00024 | $0.00404 |
| Haiku 4.5 | $0.00012 | $0.00202 |
Grade A, and why
spec-rule-miner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Spec Rule Miner
Purpose
spec-rule-miner 从目标仓库的真实代码中提炼项目级 AI 编码规则，把完整规则写入独立规则文件，并让 AGENTS.md / CLAUDE.md 这类 host 入口文件引用该文件。它是 standalone skill，不是 spec-* public workflow。
核心产物是 <=1000 words 的项目规则块，规则必须来自当前目标仓库证据，而不是语言默认、个人偏好或通用最佳实践。
When To Use
- 使用本 skill：用户要“分析项目风格”“学习代码规范”“生成项目规则”“挖掘编码习惯”“让 AI 像团队一样写代码”，或明确要生成
AGENTS.md/CLAUDE.md引用入口、Cursor/Qoder 规则文件。
When Not To Use
- 不使用本 skill：用户要审查当前 diff、修复代码、重构、调试、写 lint/format 配置、生成通用语言规范，或治理 confirmed team policy。
- 近邻路由：confirmed team policy governance 已退役，不再提供专用入口；代码质量评审走
spec-code-review；实际实现或修复走spec-work；创建或修改 spec-first source skill 走spec-write-skill。
Inputs
target_repo：必须是一个明确的本地目标仓库。- 当前仓库的人写源码、测试、配置和已有 agent rule 文件。
- 用户指定的输出目标；未指定时使用默认目标。
Outputs
rules_block：写入独立规则文件的纯规则正文，使用spec-rule-miner-start/spec-rule-miner-endmarkers。evidence_summary：每个规则组的代表性文件路径和样本限制；不写入规则文件，除非用户明确要求。target_files：默认独立规则文件、入口 pointer 文件与用户指定输出文件，说明 pointer 还是 inline。limitations：小样本、大仓库/多包抽样、混合语言、生成代码占比高、历史例外、冲突模式跳过、图候选未回源、refresh no-op、headless 默认写入等限制。
Hard Boundaries
- 只读目标仓库代码；不要修改业务源码、测试、构建配置或 formatter/linter 配置。
- Host-projected copies are outside this skill's rule targets；具体禁区见 Write Targets。宿主投影过期时从 source 运行
spec-first init修复。 - 写入独立规则文件和引用入口前必须 preview 规则正文和目标文件；交互可用时等待用户确认。只有用户明确要求直接写入，或宿主/调用参数明确证明当前运行是 headless/non-interactive，才使用默认目标；普通聊天里用户暂未回复不能算 headless。默认写入必须在 closeout 记录
headless_default_write、目标文件和限制。 - 不覆盖用户已有规则。读取目标文件后，只替换
spec-rule-minermarkers 内的旧块；无 markers 时追加；疑似旧版无 marker 输出时先询问替换还是追加。 - 非首次执行必须先重新取证并生成 candidate rules block，再与现有 canonical marked block / pointer 对比；无实质变化时不重写文件，closeout 记录
refresh_noop、采样范围和限制；有变化时 preview diff 后只替换 marker 内内容。 - 每条规则必须有当前仓库证据：默认至少 2 个文件支撑；小仓库样本不足时降级说明 sample-size；不确定或 50/50 分裂的模式不写成规则。
- 不泄露敏感信息：密钥、内部 URL、私有包名、账号、生产路径、安全实现细节只用于判断，不进入规则正文。
Workflow
- 明确
target_repo。父级多仓工作区必须先锁定一个目标仓库；不清楚时只问一个问题。 - 盘点仓库形态：根目录、主要语言、源码目录、测试目录、配置文件、包/应用边界、生成物/依赖目录和已有 agent rule 文件；多包 workspace 必须识别具体子项目范围。
- 过滤读取范围：跳过依赖、构建产物、锁文件、minified 文件、二进制、vendored/generated 代码；大仓库或多包仓库使用分层抽样并在 preview 和 closeout 中披露样本包、未覆盖子项目和适用范围。
- 读取并记录证据。抽取前先读 Pattern Categories，用其中类别组织证据；大仓库可按其中 capability-class 边界使用
code-graph/project-graph候选缩小阅读范围，但规则证据必须回到当前源码；配置已强制的 formatter/linter 规则只记录为“已由工具处理”，不要重复写入 AI 规则。 - 合成规则：按
frequency x deviation from defaults排序，保留高频且偏离默认的做法；多包规则必须区分跨包通用模式、包级专属模式和历史例外，旧项目反例只能写成“新增代码优先”或“不扩大例外”；除非证据在适用范围内压倒性一致，不使用全仓库绝对措辞。必须包含至少一个 hidden association 和至少一个 anti-pattern，除非证据明确不存在，并在 preview 限制中说明。规则正文可以包含适用范围和例外边界，这不算挖掘过程元说明。 - Preview：展示将写入独立规则文件的规则块、入口引用文件、word count、采样/证据限制、适用包范围、历史例外、refresh diff/no-op 判断，以及每个规则组的代表性 source refs。规则正文不要包含挖掘过程元说明。
- 写入前读 Write Targets，按目标文件的 marker、frontmatter、pointer/inline 规则执行。默认把完整规则写入
docs/ai/project-rules.md，并让AGENTS.md与CLAUDE.md指向该文件。 - 收尾输出：列出写入文件、规则字数、是否采样、未写入的近邻工具文件、需要用户手动检查的限制；如果没有写文件，说明 preview-only 状态；如果因 headless 走默认写入，必须说明
headless_default_write的证据来源。
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 74 lines · 118 tokens per session scan A 01b5af41925d
spec-rule-miner is a skill published in the GitHub repository ronronner02/codepilot-agent (0 stars, last pushed 3d ago), licensed MIT. It adds 118 tokens to every session and 2,020 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
agent-orchestration
Agent orchestration patterns for agentic loops, multi-agent coordination, alternative frameworks, and multi-scenario workflows. Use when building autonomous agent loops, coordinating multiple agents, evaluating CrewAI/AutoGen/Swarm, or orchestrating complex multi-step scenarios.
generate-ai-rules
Generate AI assistant configuration files for a repository — CLAUDE.md, AGENTS.md, and Cursor rules (.cursor/rules/.mdc) — from codebase analysis. Use whenever the user wants to create or update CLAUDE.md, AGENTS.md, agent rules, Cursor rules, AI coding assistant configuration, or "onboard AI tools" to a project, even…
generate-readme
Generate or refresh a comprehensive, professional README.md for a repository, with architecture overview, mermaid and optional C4 diagrams, repository structure, dependencies, and API documentation. Use whenever the user asks to create, write, update, improve, or regenerate a README, project documentation, or a…
analyze-codebase
Run a multi-agent deep analysis of a codebase, producing AI-readable analysis documents in .ai/docs/ covering structure, dependencies, data flow, request flow, and APIs. Use whenever the user asks to analyze a repository, generate codebase analysis, understand an unfamiliar codebase in depth, or before generating…
langgraph-orchestration
Build or modify LangGraph StateGraph agents in vpeetla-ai repos: typed state, nodes, conditional edges, MemorySaver, interruptbefore HITL. Use when adding orchestrators, coding loops, or multi-agent graphs.
aegis-gateway
Integrate AegisAI gateway before tool side effects (notify, publish, deploy). Use when adding Slack/Telegram/WhatsApp notify, content publish, or any irreversible external action in VAP, AegisLoop, or ai-content-factory.