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/guilopes29/brain-mcp/claude-mdgit clone --depth 1 https://github.com/GuiLopes29/brain-mcpWrote 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/guilopes29/brain-mcp/claude-md)<a href="https://agentmods.dev/instructions/guilopes29/brain-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/guilopes29/brain-mcp/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 | $0.01169 | $0.01169 |
| Opus 5 | $0.00584 | $0.00584 |
| Sonnet 5 | $0.00234 | $0.00234 |
| Haiku 4.5 | $0.00117 | $0.00117 |
Grade A, and why
brain-mcp 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 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.
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 — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Brain MCP
Semantic memory server for Claude and Cursor. Stores knowledge, decisions, and learned solutions as vector embeddings, searchable via natural language.
Architecture
packages/mcp-server/ — Node.js MCP server (stdio) + HTTP bridge (Express :3456)
packages/brain-ui/ — React + Vite UI (force-directed knowledge graph, :5173)
SQLite — Metadata + vectors (sqlite-vec), packages/mcp-server/data/brain.db
No Docker, no local Ollama — vectors live in the same SQLite file as everything else via sqlite-vec, and embeddings run in-process via @huggingface/transformers (Transformers.js).
Starting the stack
# 1. Start HTTP bridge (for the UI) — first run downloads the embedding model (ONNX, cached locally)
pnpm start:api # http://127.0.0.1:3456
# 2. Start UI (optional)
pnpm dev:ui # http://localhost:5173
MCP Tools
| Tool | Purpose |
|---|---|
get_guidelines |
Token-cheap. Compact rules + pitfalls for a project (one-line directives). Call once at task start. |
add_knowledge |
Store a solution, decision, or learning (with kind + directive) |
search_knowledge |
Semantic search across all stored knowledge |
list_knowledge |
List items, with optional project/tag filters |
update_knowledge |
Edit an existing item (re-embeds automatically) |
delete_knowledge |
Permanently remove an item by id |
All tools accept a source field (claude / cursor / browser / manual) used by the Control Room dashboard for attribution.
Active-improvement loop (token-controlled)
The Brain is not just storage — it actively shapes how AIs work:
- At task start, the consuming AI calls
get_guidelines(project)ONCE → gets ≤12 one-line directives (rules to follow, pitfalls to avoid). Cheap (~hundreds of tokens), not thousands. - On solving something, it stores knowledge with a
kind(solution/rule/pitfall/decision) and a one-linedirective. - Next session, those directives surface via
get_guidelines→ the lesson is reinforced, the vício is avoided.
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.
- 4d ago First seen · 95 lines · 1,169 tokens per session scan A e8773ab2fed7
brain-mcp CLAUDE.md is an instructions file published in the GitHub repository GuiLopes29/brain-mcp (0 stars, last pushed 1mo ago), licensed MIT. It adds 1,169 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 2026-08-31.
Other instructions, from other repositories
cognirepo CLAUDE.md
Claude Code instructions for ashlesh-t/cognirepo, covering claude.md, key rules, session start sequence (run in this order), behavioral confirmation rule and personas (cognirepo-402, cognirepo-403).
aeon CLAUDE.md
Claude Code instructions for aeonfun/aeon, covering aeon, how aeon works, strategy, voice and soul file hierarchy (read in this order).
wayland-core copilot-instructions.md
Copilot instructions for FerroxLabs/wayland-core, covering ijfw rules, output discipline, memory routing, context discipline and cross-audit.
mcp-structured-memory CLAUDE.md
Claude Code instructions for nmeierpolys/mcp-structured-memory, a project described as: Structured Memory MCP Server.
inkwell-memory CLAUDE.md
Instructions for veronchenko/inkwell-memory, covering claude.md — inkwellmemory, layout, multi-tenant mode (inkwellmultitenant=1), conventions and testing.
memory-mcp-1file AGENTS.md
Instructions for pomazanbohdan/memory-mcp-1file, covering ⛔ l0 invariants (never violate under any circumstance), detection heuristic, 🔀 phase transition routing (⛔ blocking), gate-0: phase identification & loading and 🧾 proof-of-load requirement (critical).