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/vima-tech/innate/agents-mdgit clone --depth 1 https://github.com/vima-tech/InnateWhat 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 | $0.04566 | $0.04566 |
| Opus 5 | $0.02283 | $0.02283 |
| Sonnet 5 | $0.00913 | $0.00913 |
| Haiku 4.5 | $0.00457 | $0.00457 |
Grade A, and why
Innate 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 — 252 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
This file is the single source of truth for AI coding agents (Claude Code, Codex,
opencode, Gemini CLI, …) working in this repository. Tool-specific shells such as
CLAUDE.md import this file via @AGENTS.md — keep all shared guidance here.
Editing rule (applies to all agents): Any request to "update / improve
CLAUDE.md" or the project guidance means editAGENTS.md.CLAUDE.mdis a generated shell that only contains@AGENTS.md— never add real guidance to it. Other tool shells (GEMINI.md, etc.) follow the same rule.
Commands
# Build the binary
cd core && cargo build --release
# Binary: core/target/release/innate
# Run tests
cd core && cargo test
# Recall-quality eval suite (regression safety net for fused-score / ACT-R tuning)
cd core && cargo test --release eval -- --nocapture
# Run the CLI (after adding to PATH or using full path)
innate recall "query" --format json
innate inspect
innate evolve --trigger manual
# Start MCP server (for Claude Code / Claude Desktop integration)
innate mcp
# Daemon (Linux only) — background log/hook watcher
innate daemon start --watch /path/to/log/dir
innate daemon status
innate daemon stop
MCP Integration
Add to .claude/settings.json to enable MCP tools directly in Claude Code:
{
"mcpServers": {
"innate": {
"command": "/path/to/core/target/release/innate",
"args": ["mcp"]
}
}
}
Authoritative Design Reference
docs/Innate-设计文档-v0.1.9.md is the 编码基线 (latest; supersedes the v0.1.8 doc after the modular refactor). Every design decision references a section. When behavior is ambiguous, consult the doc first.
Architecture
Conceptual model — 记忆 · 技能 · 直觉 (Memory · Skill · Intuition)
Innate presents as three cooperating layers over one procedural-knowledge core. Keep this framing when writing user-facing copy, and map each layer to its real mechanism:
| Layer | Mechanism in code |
|---|---|
| 记忆 Memory | recall → record → evolve flywheel; confidence EMA + decay + curate (kb/recall.rs, kb/record/, kb/evolve.rs, kb/curate.rs) |
| 技能 Skill | kind="skill" / origin="installed" chunks; innate-memory SKILL.md; install/ wizard |
| 直觉 Intuition | appraise critic — synchronous, no-LLM, value-domain-safe (no answer text) (kb/appraise.rs, innate_appraise) |
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 · 252 lines · 4,566 tokens per session scan A 09d207acd7a0
Innate AGENTS.md is an instructions file published in the GitHub repository vima-tech/Innate (2 stars, last pushed 25d ago), licensed MIT. It adds 4,566 tokens to every session, about $0.0228 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
agentic-context-engine AGENTS.md
Instructions for kayba-ai/agentic-context-engine, covering agents.md, repository guidelines, pipeline-first development (mandatory), core code protection and documentation maintenance.
graymatter AGENTS.md
Instructions for angelnicolasc/graymatter, covering agents.md, your tools, when to call which, first call and identity.
graymatter CLAUDE.md
Instructions for angelnicolasc/graymatter, covering claude.md, a user asked you to set graymatter up, you are working in a project that already uses graymatter and you are contributing to graymatter itself.
mnemon AGENTS.md
Instructions for mnemon-dev/mnemon, covering mnemon agent guidelines, development, go engineering and commit discipline.
mnemon CLAUDE.md
Instructions for mnemon-dev/mnemon, covering mnemon — project guidelines and development.
amfs CLAUDE.md
Instructions for raia-live/amfs, covering amfs memory — agent instructions, available mcp tools, identity, brain tools (agent-scoped) and shared knowledge tools.