ESP-Claw is an AI agent framework that runs on Espressif IoT chips and lets people define device behavior through conversation. It handles local sensing, decision-making, and execution for connected devices, with support for event-driven actions, structured memory, and MCP communication.
Borrowing it
Nothing to install: this file belongs to espressif/esp-claw. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/espressif/esp-claw/master/AGENTS.mdgit clone --depth 1 https://github.com/espressif/esp-clawWrote 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/espressif/esp-claw/agents-md)<a href="https://agentmods.dev/instructions/espressif/esp-claw/agents-md"><img src="https://agentmods.dev/badge/instructions/espressif/esp-claw/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/espressif/esp-claw/agents-md"><img src="https://agentmods.dev/badge/instructions/espressif/esp-claw/agents-md.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.02118 | $0.02118 |
| Opus 5.5 | $0.00847 | $0.00847 |
| Sonnet 5.5 | $0.00424 | $0.00424 |
| Haiku 4.5 | $0.00212 | $0.00212 |
Grade A, and why
esp-claw 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 18d 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 — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agents.md
This file provides guidance to agents when working with code in this repository.
Project Overview
ESP-Claw is an ESP-IDF firmware project for running an AI agent framework on Espressif IoT devices. The main application is application/edge_agent/; reusable firmware components live under components/. The repo also contains board definitions, build-time FATFS content, documentation, and the embedded device settings UI.
Development Commands
Export ESP-IDF before firmware work:
. $IDF_PATH/export.sh
Generate board manager files and build from the app directory:
cd application/edge_agent
idf.py bmgr -c ./boards -b esp32_S3_DevKitC_1
idf.py build
idf.py flash monitor
Docs site:
cd docs
pnpm install
pnpm build
pnpm dev
Embedded settings UI:
cd application/edge_agent/components/http_server/frontend_source
pnpm build
pnpm typecheck
High-Level Architecture
Boot and Runtime Flow
The main entry point is application/edge_agent/main/main.c.
Core Data Flow
- IM channels, scheduler jobs, Lua scripts, startup hooks, or CLI commands publish events or submit requests.
claw_event_routermatches events against the DATA root'srouter_rules/router_rules.jsonand can call capabilities, run scripts, run the agent, send messages, emit events, or drop events.claw_corebuilds context from memory, session history, skills, and other providers; calls the configured LLM backend; executes capability tool calls; persists context; and returns responses.- Outbound messages are routed back through registered IM bindings or local/web channels.
Key Subsystems
- Application shell (
application/edge_agent/main/main.c,components/common/app_claw/): boot flow, storage paths, capability registration, Lua module registration, CLI, and agent startup. - Agent core (
components/claw_modules/claw_core/): request queue, context building, LLM backend runtime, tool-call loop, media inference, interrupts, context persistence, and response delivery. - Event router (
components/claw_modules/claw_event_router/): declarative event routing and actions backed by router rules in FATFS. - Capability registry (
components/claw_modules/claw_cap/): common registration and dispatch layer for model-callable capabilities. - Capabilities (
components/claw_capabilities/): concrete agent capabilities such as Lua execution, files, IM platforms, MCP, skill management, router management, scheduler, session management, time, HTTP requests, web search, system, and LLM inspection. - Memory (
components/claw_modules/claw_memory/): session history, profile/long-term memory providers, memory persistence, request gating, and stage notes. - Skills (
components/claw_modules/claw_skill/, componentskills/directories): user-facing skill documents and activation state. - Apps (
components/common/app_registry/, SYSTEM/DATAapps/directories): standalone Lua App packages consumed by the System UI Launcher. - Lua modules (
components/lua_modules/): Lua drivers and higher-level modules for hardware, media, HTTP server, storage, threading, JSON, board manager, and capability calls. - Board manager (
application/edge_agent/boards/): board metadata, peripheral YAML, board setup code, board defaults, optional local components, and optional board FATFS overlays. - FATFS images (
application/edge_agent/fatfs_image/): build-time source trees for the read-only SYSTEM image and writable DATA seed image. - HTTP config service (
application/edge_agent/components/http_server/): local device configuration server and embedded frontend.
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.
- 18d ago Changed · +4 lines · +140 tokens per session 8be146fd88a2
- 2mo ago First seen · 150 lines · 1,978 tokens per session scan A 901719dab3af
esp-claw AGENTS.md is an instructions file published in the GitHub repository espressif/esp-claw (2,352 stars, last pushed 5d ago), licensed Apache-2.0. It adds 2,118 tokens to every session, about $0.0085 per session on Opus 5.5. 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.
esp-claw AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| esp-claw AGENTS.mdespressif/esp-claw | Instructions file | 2.4k | 5d ago | Apache-2.0 | A | 2,118 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | yesterday | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |