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/nedonion/lobster0/agents-mdgit clone --depth 1 https://github.com/NEDONION/lobster0Wrote 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/nedonion/lobster0/agents-md)<a href="https://agentmods.dev/instructions/nedonion/lobster0/agents-md"><img src="https://agentmods.dev/badge/instructions/nedonion/lobster0/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 | $0.02345 | $0.02345 |
| Opus 5 | $0.01172 | $0.01172 |
| Sonnet 5 | $0.00469 | $0.00469 |
| Haiku 4.5 | $0.00234 | $0.00234 |
Grade A, and why
lobster0 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 — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Lobster0 Agent 工程规范
适用范围与优先级
- 本文件适用于整个仓库。
- 优先遵循系统、开发者和用户的直接指令;子目录更具体的
AGENTS.md只覆盖对应目录。 - 只修改当前任务需要的文件,交付范围最小、行为正确且经过验证的变更。
- 不把 PRD 中的规划描述成已实现功能。
项目事实
- Python 最低版本为 3.12,使用
src布局,主包位于src/lobster0/。 - 项目、仓库、Python 导入包和 CLI 使用
lobster0,产品名写作Lobster0。 - 构建和工具配置以
pyproject.toml为准,包管理优先使用uv。 - 测试位于
tests/,使用标准库unittest;静态检查使用 Ruff,行宽为 100。 - 产品范围以
docs/product/20260807_产品需求文档.md为准,架构边界以docs/architecture/20260807_系统架构.md为准。
常用命令
# 首次安装或依赖变化后
uv sync --extra dev
# 运行全部测试
uv run python -m unittest discover -s tests -v
# 静态检查
uv run ruff check .
# CLI 冒烟验证
uv run lobster0 --version
# 三平台 versioned Channel gate 与 20 轮稳定性门禁
uv run lobster0 eval run --suite channel --root evals/scenarios
uv run lobster0 eval run --suite channel --repeat 20 --json --root evals/scenarios
# Phase 6 Automation 15-case 与 20 轮稳定性门禁
uv run lobster0 eval run --suite automation --root evals/scenarios
uv run lobster0 eval run --suite automation --repeat 20 --json --root evals/scenarios
# Phase 6.5 Browser 18-case 与 20 轮稳定性门禁
uv run lobster0 eval run --suite browser --root evals/scenarios
uv run lobster0 eval run --suite browser --repeat 20 --json --root evals/scenarios
# 当前发布文档、链接、Mermaid 与 HTML 一致性
uv run python scripts/validate_docs.py
- 优先使用项目
.venv或uv run,不要假设系统 Python 的依赖状态。 - 单元测试必须离线、快速、可重复,不调用真实模型或飞书接口。
- 先运行与改动最相关的测试,再运行全部检查。
修改前要求
- 先运行
git status --short,识别并保护用户已有的未提交修改。 - 阅读相关实现、测试、
pyproject.toml和对应文档,沿用现有模式。 - 明确预期行为、边界条件和验证方式;修复缺陷时先写回归测试。
- 产品语义不明确时先从 PRD、测试和公开接口获取证据,仍会改变范围时再询问用户。
Python 代码规范
- 新增或修改的公共函数、方法、类属性和返回值必须有准确类型标注。
- 使用 Python 3.12 原生类型写法,避免无意义的
Any。 - 每个新增或修改的顶层函数、方法和类必须提供中文 docstring,说明用途、参数、返回值与实际异常。
- 一个函数只承担一个清晰职责;不要为单一实现创建接口、工厂或无用配置层。
- 优先使用标准库;新增第三方依赖前必须证明标准库和现有依赖无法满足需求。
- 模块导入时不得发起网络请求、创建运行目录、连接数据库或启动服务。
- 路径操作使用
pathlib.Path,文本文件显式使用 UTF-8,时间使用带时区的 UTC。 - 库代码使用日志而不是
print();print()仅用于明确的 CLI 输出。
架构边界
channels:飞书、Telegram、Discord 等外部消息与内部消息互转,不包含 Agent 推理逻辑;三个平台共享一个 AgentRuntime,但 Transport、Delivery、Manager、queue 与运行期故障状态相互隔离。agent:管理模型与工具循环,不知道消息来自哪个渠道。providers:封装模型协议,v0.1 只实现一个 OpenAI-compatible Provider。tools:声明并执行文件、HTTP 和受限命令;不得绕过 Policy。policy:统一处理 Workspace、命令白名单、风险和审批。storage:SQLite 结构化记录以及 Markdown 记忆、Skills 的读写边界。evolution:反馈、评测和版本化提案;不得自动修改或部署 Python 源码。- 这些目录只在对应里程碑实现时创建,不提前搭空抽象。
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 · 151 lines · 2,345 tokens per session scan A c23ed3e785ac
lobster0 AGENTS.md is an instructions file published in the GitHub repository NEDONION/lobster0 (103 stars, last pushed 4d ago), licensed MIT. It adds 2,345 tokens to every session, about $0.0117 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
dograh AGENTS.md
AGENTS.md instructions for dograh-hq/dograh, covering dograh - project overview, project structure, tech stack, local development and environment configuration.
n8n CLAUDE.md
Claude Code instructions for n8n-io/n8n, a project described as: Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
meme-search CLAUDE.md
Instructions for neonwatty/meme-search, covering claude.md, 📁 file organization guidelines, 🤖 task agent usage guidelines, when to use task agents and documentation lookup pattern.
commonly CLAUDE.md
Claude Code instructions for Team-Commonly/commonly, covering claude.md / agents.md, 🧠 product vision & architecture philosophy, what commonly is, the architecture model and key concepts.
1flowbase AGENTS.md
AGENTS.md instructions for taichuy/1flowbase, covering 记忆, 用户偏好, 本项目相关skill在, 本项目 skills and 质量控制.
VaultysClaw agent-controller.instructions.md
Use when working on the agent-controller package: tools, skills, agent runtime, WebSocket client, local SQLite DB, LLM integration, or CLI modes.