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/fmarzochi/egc/claude-mdgit clone --depth 1 https://github.com/Fmarzochi/EGCWrote 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/fmarzochi/egc/claude-md)<a href="https://agentmods.dev/instructions/fmarzochi/egc/claude-md"><img src="https://agentmods.dev/badge/instructions/fmarzochi/egc/claude-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 5.1 | $0.01164 | $0.01164 |
| Opus 5 | $0.00582 | $0.00582 |
| Sonnet 5 | $0.00233 | $0.00233 |
| Haiku 4.5 | $0.00116 | $0.00116 |
Grade A, and why
EGC CLAUDE.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 — 115 lines — stays where its author put it; the contents beside it link to each section on GitHub.
EGC: Session Memory Protocol
This project has persistent cross-session memory via the egc-memory MCP server.
At the start of every session
Call get_state with no arguments: it uses the current working directory automatically:
get_state({})
If the AI is running from outside the project directory, pass the path explicitly:
get_state({ project_path: "/absolute/path/to/this/project" })
Read the returned Markdown. It contains the decisions already made, what failed, coding preferences, and what to pick up next. Do not ask the user to re-explain any of that.
At the end of every session
COMPACTION RULE: On
/compact, callupdate_stateto persist session state. Do not create compaction-summary files anywhere — not in the project, not in any external vault or notebook.update_stateupdates the existing project state file and may initialize it if it does not yet exist.
Call update_state with a summary of this session:
update_state({
project_path: "/absolute/path/to/project",
context: "One sentence: what this project is and its current phase.",
decisions: [
{ what: "What was decided", why: "Why" }
],
avoid: [
{ what: "What failed or was rejected", why: "Why to skip it next time" }
],
preferences: [
"Coding style or workflow preference discovered this session"
],
next: [
"First thing to pick up in the next session"
]
})
update_state merges with existing state: it does not erase previous memory. Only include fields that changed this session. Leave out fields with nothing new.
Pass scope: "global" to write to the user-wide memory shared across all projects; use it only for transversal preferences and lessons, never for project-specific state. get_state appends that global memory automatically, with project and branch entries taking precedence.
Parallel sessions
When more than one session works on this project at the same time, coordinate through the session bus: call session_announce at start (registers presence and territory, doubles as heartbeat), check session_peers before picking work, and claim_path before editing shared files. A refused claim means another live session holds that path: coordinate or work elsewhere, never retry in a loop.
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 · 115 lines · 1,164 tokens per session scan A a3bc736a6703
EGC CLAUDE.md is an instructions file published in the GitHub repository Fmarzochi/EGC (48 stars, last pushed today), licensed Apache-2.0. It adds 1,164 tokens to every session, about $0.0058 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 2026-09-03.
Other instructions, from other repositories
brain-os AGENTS.md
Instructions for brainOS-HQ/brain-os, covering agents.md, context, the hard rule, tool routing (in order) and mutation safety.
memorix CLAUDE.md
Claude Code instructions for Tibu142/memorix, covering memorix — agent instructions for claude code, when to search memory (memorixsearch), when to store memory (memorixstore), when to check retention (memorixretention) and best practices.
kleosrules AGENTS.md
AGENTS.md instructions for kleosr/kleosrules, covering agents.md — repository agent handbook, law, pack, skills (on demand) and workflows.
knowit AGENTS.md
Instructions for ismaelkedir/knowit: This project uses Knowit as the default persistent knowledge base for AI coding agents.
Norn CLAUDE.md
Instructions for samad001z/Norn, covering norn — project context, what this is, stack (v1), v1 scope — build only this and not in v1 (resist).
memex CLAUDE.md
Instructions for iamtouchskyer/memex, covering memex — agent working instructions, what is this, before you touch anything, how the code is layered and files you'll actually need.