nobrainr CLAUDE.md

Instructions for using nobrainr, a shared memory service for AI coding agents. It lets agents save knowledge, search it later, and explore links between related information.

In plain words
What is it for?
It is for storing and finding information about architecture, debugging, deployment, security, frontend and backend work, documentation, and other project topics.
Why use it?
It helps agents reuse lessons from earlier work instead of rediscovering the same project details in every session.

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/vicquick/nobrainr/claude-md
Clone the repo
git clone --depth 1 https://github.com/vicquick/nobrainr
Per session 7,057 This file is loaded in full into every session.
When invoked 7,057 The same file — it is already loaded in full.
Security scan A 0 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 $0.07057 $0.07057
Opus 5 $0.03528 $0.03528
Sonnet 5 $0.01411 $0.01411
Haiku 4.5 $0.00706 $0.00706

Measured 2d ago against content hash 3baab2042990, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

nobrainr 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.

CLAUDE.md · 476 lines

How it starts

The opening of the file, as written. The whole thing — 476 lines — stays where its author put it; the contents beside it link to each section on GitHub.

nobrainr — Collective Agent Memory Service

What This Is

Persistent shared memory for AI coding agents. If you're an agent reading this: you can store learnings with memory_store, search past knowledge with memory_search, and explore the knowledge graph with entity_search / entity_graph. Everything you store is available to every other agent instance connected to this server.

Provides hybrid search (vector + full-text RRF) by default, context-enriched embeddings, chunked document ingestion with overlapping context, optional cross-encoder reranking, chunk-aware retrieval, automatic entity extraction, on-write dedup, embedding versioning, and a Vue 3 dashboard with interactive graph visualization.

Canonical Categories

Use one of these when storing memories — freeform categories are auto-normalized: architecture, debugging, deployment, infrastructure, patterns, tooling, security, frontend, backend, data, business, documentation, session-log, insight, _archived

The system learns autonomously: it summarizes, consolidates duplicates, synthesizes cross-entity insights, detects contradictions, validates its own extractions, discovers cross-machine patterns, and archives stale knowledge — all on scheduled LLM-powered jobs.

Architecture

  • Backend — Python ASGI: FastMCP (HTTP + SSE) + pure JSON API (Starlette)
  • Frontend — Vue 3 + Vuetify + Cytoscape.js (separate container, nginx)
  • PostgreSQL 18 + pgvector — storage, vector similarity, knowledge graph
  • llama-swap (Coolify-deployed) — hosts three on-GPU llama-server processes serving the entire LLM stack from a single container:
    • Main LLMQwen3.6-27B-IQ4_XS.gguf (port 5803, 32K ctx, parallel=2, all layers on GPU) handles extraction, chat, scheduler jobs, dedup, synthesis
    • EmbeddingsQwen3-Embedding-0.6B-Q8_0.gguf (port 5802, 4K ctx, all layers on GPU) 1024-dim vectors, OpenAI-compatible /v1/embeddings
    • Rerankerbge-reranker-v2-m3-Q8_0.gguf (port 5800, 4K ctx, all layers on GPU) cross-encoder rerank via llama.cpp /v1/rerank, sigmoid-normalised
  • Crawl4AI — web crawling service (CPU-only container on mcp network, port 11235)

Read the full file on GitHub · 476 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. 2d ago First seen · 476 lines · 7,057 tokens per session scan A 3baab2042990

Subscribe to this mod's changes

nobrainr CLAUDE.md is an instructions file published in the GitHub repository vicquick/nobrainr (1 stars, last pushed 12d ago), licensed MIT. It adds 7,057 tokens to every session, about $0.0353 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.