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 instructions/ivyyang1999/swob/agents-mdgit clone --depth 1 https://github.com/IvyYang1999/swobWrote 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/instructions/ivyyang1999/swob/agents-md)<a href="https://agentmods.dev/instructions/ivyyang1999/swob/agents-md"><img src="https://agentmods.dev/badge/instructions/ivyyang1999/swob/agents-md.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 | $0.01764 | $0.01764 |
| Opus 5 | $0.00882 | $0.00882 |
| Sonnet 5 | $0.00353 | $0.00353 |
| Haiku 4.5 | $0.00176 | $0.00176 |
Grade A, and why
swob AGENTS.md 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 5d 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 — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Swob 研发协作唯一规范
本文件是仓库内 Agent 协作、验收、合并与发布的唯一流程权威。CLAUDE.md、docs/PROJECT.md、docs/PARALLEL.md 只能引用或补充事实,不得定义冲突流程。
不可越过的边界
- Worker 只能在独立 worktree 和任务分支工作;不得直接修改
master,不得 push。 - 一个工作包只有一个 Owner。Builder 完成实现与本地证据,但 Builder 不得充当该工作包的 Verifier。
- 只有 Integrator 可以把已验收的工作包送入集成分支;语义冲突必须暂停并交回 Integrator/人类,不得自动猜测解决。
- 只有 yyt 可以授权 push、部署与公开发布。Agent、脚本和 Git hook 均不得自行推断授权或自动 push、部署、替换、重启应用;当前批次得到 yyt 明确授权时,Integrator 才可代为执行该批 push/本地部署,tag、Release 与官网发布仍须逐项授权。
- 每个独立改动单独提交;提交前运行与风险相称的验证。不得混入其他工作包或用户未提交改动。
角色与职责
| 角色 | 默认承担者 | 唯一职责 | 禁止事项 |
|---|---|---|---|
| Product Owner | yyt | 定义目标、优先级、接受标准与发布决定 | 不把含糊目标直接下放为实现 |
| Fable | Fable | 仅在 S2/S3 做契约与任务图、语义冲突裁决、最终 Judge | 不承担普通调研、实现或机械合并 |
| Mapper | Codex | 只读勘察代码、依赖、风险、测试与事实边界 | 不在勘察阶段改代码 |
| Contract Owner | 对目标领域最强的模型/人 | 固化接口、数据与行为契约 | 不同时分配重叠 Owner |
| Builder | Codex 或 Opus,按工作包选择 | 在独立 worktree 实现一个明确工作包并提交证据 | 不提前集成,不验证自己的高风险产出 |
| Verifier | 独立上下文;高风险时换模型或人 | 按验收标准复核结果与证据 | 不以 Builder 自证代替独立验收 |
| Integrator | Codex + Merge Queue | 只接收合格 result.json,顺序合并、运行门禁、报告冲突;有当批明确授权时执行 push/本地部署 |
不自动解决语义冲突,不把一次授权外推到后续批次或公开发布 |
| Release | yyt | 审阅集成结论,授权 push/部署,并逐项决定 tag、Release 与官网发布 | 不把发布授权隐含给 Agent 或 hook |
| Learning Miner | GLM | 从已完成批次提炼可复用经验 | 不改变当批实现或验收结论 |
生命周期
- S0:需求目标、成功标准或边界仍不清楚;先由 Product Owner 澄清。
- S1:边界清楚、低耦合;Mapper 可直接形成独立工作包。
- S2:跨模块或契约变化;Fable 先产出契约与依赖图,再分配 Builder。
- S3:高风险、语义冲突或发布决策;Fable/Judge 与人类介入。
工作包按 mapped → contracted → claimed → built → verified → integrated → releasable 单向推进;失败退回产生失败的阶段,不靠跳过门禁推进。
分支、worktree 与提交
-
从最新的受信任基线创建任务分支和独立 worktree;一个 worktree 只承载一个工作包。
-
开工前记录
baseSha、Owner、write_scope、验收证据和依赖;写入范围交集在开工前消解。 -
Worker 只提交到自己的任务分支。每个 Agent 提交必须包含以下 trailers,未知值写
n/a,不得省略:Agent-Task: <work-item-id> Agent-Harness: <codex|claude-code|other> Agent-Model: <model> Agent-Session: <session-id> Agent-Decision: <关键取舍,单行> Agent-Limitation: <已知限制,单行;无则 none> -
Worker 完成后只提交
result.json和分支/commit 定位,不 push。Verifier 独立复核后,Integrator 才可运行 Merge Queue。 -
Merge Queue 只能生成“可 push”或“不可 push/需处理”的结论,绝不执行 push/deploy。Integrator 复核报告后,只有在 yyt 对当前批次已有明确授权时才能代执行 push/本地部署;否则停在“可 push”等待授权。tag、Release 与官网发布必须逐项授权。
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.
- 5d ago First seen · 97 lines · 1,764 tokens per session scan A 91db9726ba24
swob AGENTS.md is an instructions file published in the GitHub repository IvyYang1999/swob (38 stars, last pushed 21d ago), licensed Apache-2.0. It adds 1,764 tokens to every session, about $0.0088 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 2026-08-30.
Other instructions, from other repositories
cursor-studio copilot-instructions.md
Copilot instructions for DearLicy/cursor-studio, a project described as: Local-first Windows control center for Cursor: manage AI providers, balances, usage and costs, sessions, prompts, MCP, Skills, appearance, and integrations. | Cursor 本地桌面管理工具.
opencove AGENTS.md
AGENTS.md instructions for DeadWaveWave/opencove, covering system prompt: opencove project ai developer agent, 1. core directives & golden rules, 2. decision framework (small vs large), a. small change (fast feedback / 小步快反馈) and b. large change (deep thinking / 慎重对齐).
Ataru AGENTS.md
Instructions for lovstudio/Ataru, covering agents.md, design system, quick rules, color palette and common patterns.
Ataru CLAUDE.md
Instructions for lovstudio/Ataru, covering claude.md, design system, quick rules, color palette and common patterns.
wesight AGENTS.md
Instructions for freestylefly/wesight, covering claude.md, build and development commands, build production bundle (typescript + vite), lint with eslint and run memory extractor tests (node.js built-in test runner).
limboo CLAUDE.md
Claude Code instructions for limboo-ai/limboo, covering claude.md, 1. what is limboo?, 2. tech stack (current), 3. project structure and the three electron contexts (critical mental model).