agentmods.dev

cerefox CLAUDE.md

cerefox CLAUDE.md is an instructions file for coding agents from fstamatelopoulos/cerefox. It costs 10,388 tokens per session, scanned B, original, Apache-2.0.

A project guide for Cerefox, a shared long-term memory system for AI agents. It stores Markdown knowledge in a database and makes it searchable through MCP and REST.

In plain words
What is it for?
Use it to understand Cerefox's architecture, technology choices, project structure, and development conventions.
Why use it?
It helps agents find useful information across separate conversations, sessions, and contexts instead of losing that knowledge over time.

Instructions file

not rated 16 10d ago B scan 10,388 tokens original Apache-2.0
View the source file fstamatelopoulos/cerefox
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/fstamatelopoulos/cerefox/claude-md
Clone the repo
git clone --depth 1 https://github.com/fstamatelopoulos/cerefox

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 cerefox CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/fstamatelopoulos/cerefox/claude-md.svg)](https://agentmods.dev/instructions/fstamatelopoulos/cerefox/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/fstamatelopoulos/cerefox/claude-md"><img src="https://agentmods.dev/badge/instructions/fstamatelopoulos/cerefox/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 10,388 This file is loaded in full into every session.
When invoked 10,388 The same file — it is already loaded in full.
Security scan B 2 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 $0.10388 $0.10388
Opus 5 $0.05194 $0.05194
Sonnet 5 $0.02078 $0.02078
Haiku 4.5 $0.01039 $0.01039

Measured 4d ago against content hash 6ac0968e90b4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

cerefox CLAUDE.md scanned grade B with 2 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

| OpenAI Codex CLI | `url` + `bearer_token_env_var` in `~/.codex/config.toml` | Direct Streamable HTTP; TOML config; **tested, working** |

Makes network callslowCapability

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

1. **AI agents / Edge Functions** — callers (GPT Actions, remote MCP, curl) present the **Cerefox access token** (`cfx_pat_…`, iter-28E) as `Authorization: Bearer`. Every data Edge Function runs `--no-verify-jwt` and val
CLAUDE.md · 486 lines

How it starts

The opening of the file, as written. The whole thing — 486 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cerefox - Project Guide

What Is This

Cerefox is a user-owned knowledge memory layer for AI agents. It stores curated Markdown documents in Supabase (Postgres + pgvector), supports hybrid search (FTS + semantic), and exposes everything via MCP and REST so any AI agent can read and write.

Cerefox is asynchronous shared memory, not a message bus. It solves the persistent context problem: knowledge written in one context is findable in any other, dissolving boundaries between agents, sessions, human and machine, and across time. It does not handle real-time agent-to-agent communication; protocols like A2A handle that. Cerefox handles persistent memory.

Single-user, open-source (Apache 2.0), designed to be cheap/free to operate. See docs/research/vision.md for the full project vision.

Tech Stack

  • Language: TypeScript via Bun (CLI, web server, local MCP server, build/release scripts, Edge Functions). The Python implementation was fully removed at v1.0.0 (iter-28G) — only the SQL schema assets under src/cerefox/db/ remain (they are not Python).
  • Database: PostgreSQL 16+ with pgvector (Supabase free tier or local Docker)
  • Embeddings: OpenAI text-embedding-3-small (768-dim, cloud API, default) or the local ONNX nomic-embed-text-v1.5 (768-dim, CEREFOX_EMBEDDER=local — Cerefox Local only, shipped at v1.0.0); Fireworks AI is roadmap (not wired); Edge Functions handle embedding server-side for agents (cloud path)
  • Web framework: Hono on Bun/Node (TypeScript), in packages/memory — served by cerefox web.
  • Web UI: React + TypeScript SPA (Mantine UI, TanStack Query, Vite); served at /app/
  • CLI: commander (TypeScript) in @cerefox/memory; resource-verb shape (v0.9.0).
  • Local MCP server: @cerefox/memory npm package (Node ≥20 / Bun ≥1.0); single artifact growing to host CLI + web server + ingestion in future iterations
  • Shared TS modules: _shared/{config,db-client,db-status,embeddings,mcp-tools}/ — imported by both Edge Functions (Deno) and local server (Node/Bun) via structural typing
  • Package management: bun workspaces for TS
  • Testing: bun test (TypeScript) — the only test runner as of v0.9.0 (pytest retired; tests/**/*.py deleted)
  • Type-checking: tsc --noEmit for TS (bun run typecheck)

Read the full file on GitHub · 486 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. 4d ago First seen · 486 lines · 10,388 tokens per session scan B 6ac0968e90b4

Subscribe to this mod's changes

cerefox CLAUDE.md is an instructions file published in the GitHub repository fstamatelopoulos/cerefox (16 stars, last pushed 10d ago), licensed Apache-2.0. It adds 10,388 tokens to every session, about $0.0519 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Memory and contextai-knowledge-basehybrid-searchmcppgvector
Related

Other instructions, from other repositories

ClawMem AGENTS.md

Instructions for yoloshii/ClawMem, covering clawmem — agent quick reference, inference at a glance, install, or step by step and memory retrieval (90/10 rule).

yoloshii/ClawMem · 7,138 tokens

mind-mem CLAUDE.md

Instructions for star-ga/mind-mem, covering mind-mem — persistent ai memory system, architecture, key components, mcp tools (97) and config.

star-ga/mind-mem · 4,812 tokens

pgmnemo AGENTS.md

Instructions for pgmnemo/pgmnemo, covering pgmnemo — agent integration guide, 1. what pgmnemo is — and the problem it solves, the problem, architectural consequences and 2. when to adopt / when not to.

pgmnemo/pgmnemo · 8,496 tokens

SecureContext CLAUDE.md

Instructions for iampantherr/SecureContext, covering securecontext — agent instructions, what this project is, directory layout, build commands and 13 mcp tools.

iampantherr/SecureContext · 1,689 tokens

subcog CLAUDE.md

Instructions for zircote/subcog, covering claude.md, project overview, key capabilities, build commands and primary development workflow.

zircote/subcog · 501 tokens

mind-mem AGENTS.md

Instructions for star-ga/mind-mem, covering mind-mem: agent instructions (auto-written), memory context, apr 21, 2026, may 2, 2026 and may 6, 2026.

star-ga/mind-mem · 1,002 tokens