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/lanseqingling/lantern/agents-mdgit clone --depth 1 https://github.com/lanseqingling/lanternWhat 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 | $0.02060 | $0.02060 |
| Opus 5 | $0.01030 | $0.01030 |
| Sonnet 5 | $0.00412 | $0.00412 |
| Haiku 4.5 | $0.00206 | $0.00206 |
Grade A, and why
lantern 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 yesterday.
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 — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Project
Lantern AI 是面向个人漫画创作者的开源漫画工作台，串联故事与资产管理、页面与画格编排、单格精修、预览与导出，并通过 MCP + Skill 接入外部 AI Agent，同时让创作者始终掌握关键创作决定。
内置 Agent 当前暂不迭代；除非任务明确要求，不扩展其能力。
本文件只提供开发 Agent 所需的仓库入口、稳定边界和交付要求。产品规则与协议由 docs/ 维护，字段和运行行为以可执行 schema 与代码为准；不要在这里复制功能清单、接口目录或阶段状态。
Commands
| 任务 | 命令 |
|---|---|
| 安装依赖 | ./lantern setup |
| 启动本地产品 | ./lantern start |
| 启动开发进程 | ./lantern dev |
| 运行时诊断 | ./lantern doctor |
| 安装外部 Agent 应用接入 | ./lantern agent:install |
| 创建一致备份 | ./lantern backup:create |
| 恢复一致备份 | ./lantern backup:restore <backup-file> |
| 类型检查 | ./lantern typecheck |
| 单元测试 | ./lantern test |
| 构建 | ./lantern build |
| 打包源码发行包 | ./lantern package:release |
按变更范围运行 package.json 中对应的专项命令；交付前至少完成相符的类型检查、测试和构建。
Repository Guide
| 路径 | 职责 |
|---|---|
apps/web/、apps/api/ |
Web 界面与 loopback API 入口。 |
packages/shared/ |
跨端协议、schema 与共享领域类型。 |
packages/editor-core/ |
与 UI 解耦的编辑 Capability、变更和快照能力。 |
packages/agent-runtime/ |
Agent 规划、上下文、工具与任务运行时。 |
packages/server/ |
持久化、对象存储和服务层业务编排。 |
packages/ 其他目录 |
可复用的领域、渲染、布局、演示和 UI 模块。 |
prisma/ |
数据模型与迁移。 |
docs/ |
产品、交互、协议和 Agent 的正式文档。 |
samples/、apps/web/public/samples/ |
显式示例作品与静态素材。 |
scripts/、tests/ |
启动、构建、打包、诊断脚本与自动化验证。根目录 lantern/lantern.cmd 只提供薄命令入口。 |
skills/ |
随 Lantern 分发的应用级 Agent Skill 源；由应用安装器部署到目标 Agent 的用户级目录，不作为仓库开发 Skill 自动发现。 |
Architecture Boundaries
- TypeScript 是应用与共享包的统一语言；共享契约只保留一个代码事实源。
- Workspace 之间只能通过已声明的
@lantern/*依赖和公开exports协作；不得跨包引用其他模块的src、内部文件或依赖根目录偶然提升的第三方包。 - Web 通过 API 访问服务端能力。路由负责请求解析、鉴权和响应边界，领域逻辑与持久化编排进入对应服务或领域包。
- SQLite 保存作品和工作流事实；Local Task Runner 只负责唤醒与进程内并发，不能成为任务或作品事实源。
- 图片等二进制内容进入用户数据目录中的本地对象存储；数据库与作品协议只保存稳定对象键和版本引用。
- 数据库、对象、配置、日志和临时文件必须通过统一 runtime paths 解析；生产数据不得写入仓库、安装目录或当前工作目录。
- 备份只包含作品事实与对象文件，不包含 Provider Key、安装令牌和日志；恢复必须先完成 manifest、SHA-256 与 SQLite 完整性校验，并在服务停止时原子替换数据库和对象目录。
- 本地 API 只监听 loopback，并通过安装级令牌映射到稳定本地用户；请求头不能切换用户身份。
- 确定性编辑通过领域 Capability 产生原子变更。组件、Agent 和 API 不各自复制参数契约、对象 ID 规则或业务默认值。
- 生产、演示和测试运行时必须明确隔离；生产失败不得静默回退到 mock 或示例数据。
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.
- yesterday First seen · 96 lines · 2,060 tokens per session scan A e821a132ac9d
lantern AGENTS.md is an instructions file published in the GitHub repository lanseqingling/lantern (5 stars, last pushed 1mo ago), licensed MIT. It adds 2,060 tokens to every session, about $0.0103 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 instructions, from other repositories
elicitly AGENTS.md
AGENTS.md instructions for elicitly/elicitly: This is the open-source Elicitly Free Edition: packages/elicitly (the local stdio MCP server published as elicitly) and packages/tools (the embeddable @elicitly/tools toolkit). The hosted Pro Edition lives in a separate private repository — nothing here may depend on it or…
agentcloud AGENTS.md
AGENTS.md instructions for premhiru/agentcloud, covering agentcloud engineering rules and this is not the next.js you know.
maple-procure CLAUDE.md
Claude Code instructions for samarpassey/maple-procure, covering mapleprocure, commands, read these when and repo conventions.
madeforai-supplychain-skill AGENTS.md
AGENTS.md instructions for amplichina/madeforai-supplychain-skill, a project described as: Give AI agents hands in the real world. An open-source MCP execution control layer between AI agents and China-based manufacturing — with humans holding every irreversible gate.
llm-context.py CLAUDE.md
Instructions for cyberchitta/llm-context.py, covering claude.md, working notes (gitignored) and draining the field notes.
seekstone CLAUDE.md
Instructions for shaqmughal/seekstone, covering claude.md, what this repo is, commands, the harness itself (run after npm install) and architecture.