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/memi-design/memi/agents-mdgit clone --depth 1 https://github.com/memi-design/memiWrote 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/memi-design/memi/agents-md)<a href="https://agentmods.dev/instructions/memi-design/memi/agents-md"><img src="https://agentmods.dev/badge/instructions/memi-design/memi/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 | $0.01752 | $0.01752 |
| Opus 5 | $0.00876 | $0.00876 |
| Sonnet 5 | $0.00350 | $0.00350 |
| Haiku 4.5 | $0.00175 | $0.00175 |
Grade A, and why
memi 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 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 — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Mémoire — Guidelines for Codex
Prime Directives
- Operate as an autonomous superagent — read designs, drive canvas, generate specs and code without waiting for permission. Burn tokens for quality. (Why: Mémoire is built for hands-free operation; hesitation breaks the flow.)
- Use Atomic Design exclusively — every component specifies a level: atom → molecule → organism → template → page. (Why: enforced in specs, codegen, and Figma structure.)
- Self-heal after every canvas operation — CREATE → SCREENSHOT → ANALYZE → FIX → VERIFY (max 3 rounds). (Why: floating elements, wrong sizing, and raw hex values are the top defects.)
- Check Code Connect before creating anything —
get_code_connect_mapfirst. Use mapped components when they exist. (Why: duplicate components are the #1 design system debt source.) - Load relevant skills before acting — skills in
skills/define how agents operate. Read the skill file that matches your task.
What is Mémoire
AI-native design intelligence engine. Bridges Figma, user research, and code generation into a spec-driven system. Auto-discovers and connects to the Figma plugin, generates shadcn/ui components from structured specs, and supports multiple Codex instances natively via box widgets.
Architecture
| Directory | Purpose |
|---|---|
src/engine/ |
Core orchestrator, project detection, registry |
src/figma/ |
Figma bridge (WebSocket auto-discovery on ports 9223-9232), tokens, stickies |
src/research/ |
Research engine (Excel, web, stickies → insights) |
src/specs/ |
Spec types (component, page, dataviz, design, ia) + Zod validation |
src/codegen/ |
Code generation (shadcn mapper, dataviz, pages) → atomic folders |
src/notes/ |
Mémoire Notes — downloadable skill packs (loader, resolver, installer) |
src/preview/ |
Localhost preview gallery (HTML + API server) |
src/agents/ |
Agent orchestrator, multi-agent support, self-healing, box widgets |
src/tui/ |
Terminal UI (Ink/React) |
src/commands/ |
CLI commands (Commander.js) |
skills/ |
Built-in skill definitions — ship with the npm package |
plugin/ |
Figma plugin (auto-discovers Mémoire on ports 9223-9232) |
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 · 108 lines · 1,752 tokens per session scan A b876ef94f5c1
memi AGENTS.md is an instructions file published in the GitHub repository memi-design/memi (40 stars, last pushed yesterday), licensed MIT. It adds 1,752 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
ui-mcp CLAUDE.md
Instructions for Forge-Space/ui-mcp, covering siza mcp, quick reference, project, architecture and stack.
typescript-sdk CLAUDE.md
Claude Code instructions for modelcontextprotocol/typescript-sdk, covering claude.md, build & test commands, run a single package script (examples), breaking changes and code style guidelines.
frontmcp CLAUDE.md
Claude Code instructions for agentfront/frontmcp, covering skills and tools, documentation site layout, frontmcp monorepo - development guide, repository structure and primary libraries (scoped under @frontmcp/).
figmosha2 CLAUDE.md
Instructions for denysosadchyi/figmosha2, covering figmosha 2.0 — claude code instructions, how to send code, preferred — subcommand-style, shorthand (auto-prepends exec) and quick http (no python needed).
prism-coder GEMINI.md
Gemini CLI instructions for dcostenco/prism-coder: See the Synalux engineering standards doc for the full protocol. This file inherits those standards. Additional Prism-specific rules.
yahoo-finance2 AGENTS.md
Instructions for gadicc/yahoo-finance2, covering agent notes, project context, commands, testing and fixtures and schema and validation.