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/letsagents/aseprite-mcp/agents-mdgit clone --depth 1 https://github.com/letsagents/aseprite-mcpWrote 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/letsagents/aseprite-mcp/agents-md)<a href="https://agentmods.dev/instructions/letsagents/aseprite-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/letsagents/aseprite-mcp/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 | $0.03031 | $0.03031 |
| Opus 5 | $0.01515 | $0.01515 |
| Sonnet 5 | $0.00606 | $0.00606 |
| Haiku 4.5 | $0.00303 | $0.00303 |
Grade A, and why
aseprite-mcp 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 2d 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 — 138 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — 跨工具项目指令（给 AI 助手）
本项目是 aseprite-mcp：一个用 TypeScript 编写的 Aseprite MCP（Model Context Protocol）server。它把 Aseprite 以 76 个工具的形式暴露给 LLM 客户端，通过生成 Lua 脚本 + headless 调用 Aseprite -b --script 来驱动 Aseprite。你（AI 助手）在这个项目里改代码时，遵守本文件。
本文件是跨工具指令（Codex / Cursor / Copilot / Gemini 等共用同一份）；Claude Code 经 CLAUDE.md 的 @AGENTS.md import 读到全文。本文件给后续 AI（或人）在本仓库内开发、维护、扩展时作指引：架构约定 + 红线 + 速查，不是教程。
常用命令
npm install # 安装依赖
npm run build # tsc -p tsconfig.build.json → dist/（不含测试）
npm run dev # tsx src/index.ts 直接运行（无需先 build）
npm start # node dist/index.js 运行编译产物
npm test # node:test 跑全部 *.test.ts
npm run typecheck # tsc --noEmit 全量类型检查
若自动探测（--aseprite-path > env ASEPRITE_PATH > PATH 搜索 > Windows 默认位置，见 src/aseprite/path.ts）找不到 Aseprite，需显式指定：
ASEPRITE_PATH="/path/to/your/Aseprite.exe" npm start
架构与数据流
一次工具调用的完整链路：
MCP 客户端请求
→ SDK 高阶 API（McpServer，zod 校验参数）
→ tools/<域>.ts 的 handler
→ lua/emit/<域>.ts 生成 Lua 字符串（纯函数，不碰 IO）
→ Ctx.run(lua)
→ aseprite/cli.ts 的 AsepriteRunner：写临时 .lua → spawn Aseprite -b --script → 收 stdout
→ classify（按 stdout 末尾 __SCRIPT_OK__/__SCRIPT_ERROR__ 标记清洗成 RunOutcome）
→ handler 把 RunOutcome 包成 ToolOutcome { isError, text }
→ SDK 回 MCP 响应
分层（依赖严格单向，自底向上）：
| 层 | 文件 | 职责 |
|---|---|---|
| 核心 | error.ts runner.ts context.ts |
错误类型、LuaRunner 接口、Ctx（runner + verbose） |
| 桥接 | aseprite/path.ts aseprite/cli.ts |
Aseprite 定位、spawn + xpcall wrapper + classify |
| Lua 工具 | lua/color.ts lua/escape.ts |
hex 解析、Lua 字符串转义 |
| Lua 生成 | lua/emit/*.ts（+ shared.ts） |
每个工具一段 Lua 代码生成（纯字符串拼接） |
| 工具注册 | tools/*.ts（+ shared.ts index.ts） |
zod schema + handler + register |
| 接线 | server.ts index.ts |
装配 McpServer（工具 + resource + prompt）、CLI、stdio |
关键约定（改代码前必读）
工具的三段式结构
tools/ 下每个域文件对每个工具导出三样（命名 <toolName> 用 snake_case 工具名转 camelCase，如 draw_circle → drawCircle）：
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.
- 2d ago First seen · 138 lines · 3,031 tokens per session scan A 55f831a990f3
aseprite-mcp AGENTS.md is an instructions file published in the GitHub repository letsagents/aseprite-mcp (0 stars, last pushed 22d ago), licensed MIT. It adds 3,031 tokens to every session, about $0.0152 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
pixellab-pip GEMINI.md
Instructions for Shilo/pixellab-pip: Use PixelLab Pip when the user asks about PixelLab asset generation, editing, animation, MCP setup, REST v2 integration, website/editor workflows, Aseprite, Pixelorama, legacy v1, bearer-token auth, endpoint choice, SDK coverage, or PixelLab docs.
pixellab-pip AGENTS.md
Instructions for Shilo/pixellab-pip: PixelLab Pip is an unofficial, agent-agnostic PixelLab workflow router shipped as a portable Agent Skill. It teaches coding agents to choose between documented PixelLab MCP tools, REST v2 endpoints, website/editor fallback, Aseprite, Pixelorama, and local asset assembly. The…
ase-deliver AGENTS.md
AGENTS.md instructions for ylty1516/ase-deliver, covering asedeliver — contract for any ai, what this tool is, do this, in order, templates and hard visual rules.
pixel-plugin CLAUDE.md
Instructions for willibrandon/pixel-plugin, covering claude.md, project overview, architecture, directory structure and plugin component specifications.
ase-deliver CLAUDE.md
Claude Code instructions for ylty1516/ase-deliver: Follow AGENTS.md. Compile sprites with python -m asedeliver. The deliverable is always out/ .aseprite.
pixel-mcp CLAUDE.md
Instructions for willibrandon/pixel-mcp, covering claude.md, project overview, build commands, build the server binary and or: go build -o bin/pixel-mcp ./cmd/pixel-mcp.