aimeat-protocol CLAUDE.md

aimeat-protocol CLAUDE.md is an instructions file for coding agents from miikkij/aimeat-protocol. It costs 10,700 tokens per session, scanned D, original, MIT.

A set of project principles for AIMEAT, an AI-centered system where chat is the main interface and MCP is the preferred connection to capabilities. It emphasizes user ownership of knowledge, identity, consent, and tracking where data comes from.

In plain words
What is it for?
Use it when proposing features, choosing between chat and web interfaces, handling user knowledge and permissions, or deciding how agents should connect to the system.
Why use it?
It helps prevent designs that force users into web forms, weaken control over their data, or treat AI agents as secondary users.

Instructions file

Install

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.

agentmods
npx agentmods add instructions/miikkij/aimeat-protocol/claude-md
Clone the repo
git clone --depth 1 https://github.com/miikkij/aimeat-protocol

Wrote 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.

agentmods badge for aimeat-protocol CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/miikkij/aimeat-protocol/claude-md.svg)](https://agentmods.dev/instructions/miikkij/aimeat-protocol/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/miikkij/aimeat-protocol/claude-md"><img src="https://agentmods.dev/badge/instructions/miikkij/aimeat-protocol/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 10,700 This file is loaded in full into every session.
When invoked 10,700 The same file — it is already loaded in full.
Security scan D 3 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.10700 $0.10700
Opus 5 $0.05350 $0.05350
Sonnet 5 $0.02140 $0.02140
Haiku 4.5 $0.01070 $0.01070

Measured yesterday against content hash e173a8674e8a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade D, and why

aimeat-protocol CLAUDE.md scanned grade D with 3 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.

Sends data to an external URLmediumData exfiltration

A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.

Tooling that has bitten before: the dev server does not watch backend `src/` (restart for a new route) · Playwright MCP needs `--isolated` and cannot use `file://` · **any** recursive delete follows a junction, `git work

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

Tooling that has bitten before: the dev server does not watch backend `src/` (restart for a new route) · Playwright MCP needs `--isolated` and cannot use `file://` · **any** recursive delete follows a junction, `git work

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- **Never claim anything about prod without probing it.** `curl /v1/build` gives the restart time (`parseInt(build,36)`), and grepping a live asset for a marker from the change is the definitive proof. Saying "you are on
CLAUDE.md · 295 lines

How it starts

The opening of the file, as written. The whole thing — 295 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md: AIMEAT Protocol

What this is for

Read this before proposing anything. When a design question is genuinely open, these decide it, and every one of them has cost a rewrite when it was forgotten.

The goal is that people become AI-native. Success is someone doing their work through their own AI and finding the system worth having. Logging into a website to type into forms is the fallback, not the destination.

AI chat is the primary interface, and MCP is the preferred road in. The web is the machine room: it shows status and holds the controls that genuinely need a screen. So a capability reachable only by clicking is not finished. Design the chat path first, then the surface that shows what happened.

The data is refined knowledge the user brought and owns. Ownership is the product. That is why identity, consent and provenance are load-bearing rather than compliance overhead, and why a feature that erodes ownership is wrong even when it is convenient.

Agents are first-class users. They arrive through device authorization, the hello-integration flow and the aimeat-crewai liaison package, and an agent should be able to do what a person can do. An agent-shaped door added after the fact is a symptom that the feature was designed for a screen.

People's intelligences come from different vendors and arrive from different directions. The system's job is to let them share what they know without a migration and without a lock-in. Organisms, workspaces and skills exist for that.

Knowledge has a lifecycle. What matters stays relevant; what stopped mattering ages, gets marked, or is cleaned away. A store with no way to forget becomes what this project's own local memory became by 2026-08-09: 1.6 MB that loaded whether or not it applied. Any surface that accumulates knowledge needs an answer to "how does this go stale".

One more, from how this is sold rather than built: lead with what a person gets, not with the protocol.


The AIMEAT protocol (AI Memory Exchange and Action Transfer) and its reference implementation. Three parts:

Read the full file on GitHub · 295 lines

Changes

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.

  1. yesterday Changed · +5 lines · +996 tokens per session e173a8674e8a
  2. 5d ago First seen · 290 lines · 9,704 tokens per session scan D 6359af63c7de

Subscribe to this mod's changes

aimeat-protocol CLAUDE.md is an instructions file published in the GitHub repository miikkij/aimeat-protocol (7 stars, last pushed yesterday), licensed MIT. It adds 10,700 tokens to every session, about $0.0535 per session on Opus 5. A static security scan graded it D with 3 findings (sends data to an external url, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.