Brain CLI — AI skill for Claude Code
Persistent memory for AI agents and personal knowledge management.
How to install Brain CLI
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open victorsabino/brain-cli and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Brain CLI does
Persistent memory for AI agents and personal knowledge management. Single-file Python CLI: hybrid semantic + keyword search (SQLite FTS5 + sqlite-vec embeddings, RRF fusion), dedup, tasks, linking. Local-first, no server. Built to be driven by Claude Code, Cursor, and other LLM coding agents.
Alternatives in AI
- Codeseek — Rust-powered code intelligence CLI for AI coding agents 764 ★
- Fusion 3 — Fusion full panel — Opus 4.8 + GPT-5.5 + Gemini 3.1 Pro in parallel, judged by Opus 4.8 (opus4.8-gpt5.5-gemini 468 ★
- Ataru — High-performance AI memory retrieval for local agent history — a Rust search core (SDK / API / JSON CLI) plus 351 ★
README
brain — local memory for AI agents, in a single Python file
[](https://github.com/victorsabino/brain-cli/actions/workflows/ci.yml)
Your coding agent forgets everything the moment the session ends. `brain` gives it a persistent local memory it can search — one SQLite file, no server, no cloud.
**Example output** (`brain search "deploy regression" --explain`):
[bug ] Rollback loop on deploy (acme)
...
2026-05-01 · ab12cd34ef56 · score=1.05 #deploy
↳ fts=#1 sem=#2 sim=0.712 rrf=0.500+0.492 rec=+0.044 acc=+0.011 = 1.047
A persistent memory / personal knowledge base CLI for LLM coding agents (Claude Code, Cursor, custom agents) and humans: save, search, link and dedupe learnings, decisions, bugs, snippets and tasks in one local SQLite file. Hybrid semantic + keyword search — porter-stemmed FTS5 plus chunked 384-dim embeddings (sqlite-vec, cosine) fused with reciprocal rank fusion (RRF). No server, no cloud, no SQL required: typed argparse commands over `~/brain.db` (override with `BRAIN_DB=/path/to.db`). The schema is additive over any existing `memories` table; no destructive migration.
Works with any agent that can run a shell command: Claude Code, Cursor, Windsurf, GitHub Copilot, OpenAI Codex CLI, Gemini CLI, Aider, Cline, Roo Code, Continue, Goose — point it at the `brain` binary and give it the save/search conventions in your rules file (CLAUDE.md, .cursorrules, AGENTS.md, etc.).
🤖 For agents (and the humans who run them)
**Agents:** read [AGENTS.md](AGENTS.md) and follow it — it's an executable runbook that walks you through installing brain, interviewing your user (install method, semantic search on/off, DB location), verifying the setup, and wiring memory conventions + an optional auto-capture hook into your own rules file. Every step that touches the user's machine asks for consent first.
**Humans:** the whole setup is
Related Skills
Anyindex MCP
Local MCP server for semantic code search. AST-aware chunking with tree-sitter, offline ONNX embeddings, hybri
Doc Search MCP
Local MCP server that indexes PDFs, EPUBs, HTML, Markdown and text files and makes them searchable by LLM codi
Guild
Shared context, memory, and task coordination across AI coding agents. Single Go binary, local SQLite, hybrid
Local Memory
100% local, OpenAI-compatible memory layer for AI agents. On-device embeddings via Ollama + SQLite (sqlite-vec
Enfold
Enfold. Every memory enfolds the whole. Local, private long-term memory for AI agents: one SQLite brain shared
Codexlens Search
Lightweight semantic code search engine — 2-stage vector + FTS + RRF fusion + MCP server for Claude Code
Related Agents
Python RAG Backend Engineer
Production-grade Python RAG backend engineer specializing in ChromaDB vector storage, sentence-transformers em
Duplication Detector
Detects stories with overlapping implementation concerns using hybrid keyword pre-filter and LLM semantic veri
Audit Cluster
Maps topic clusters, detects keyword cannibalization via SERP overlap, validates hub-spoke internal linking, i