robo-cortex — Press Kit
Git-aware knowledge integrity for AI coding agents.
About
Your AI agent debugs a bug, learns the lesson, and forgets it by next week — or worse, keeps citing a "lesson" that the code has since made false. robo-cortex is a local knowledge base built for that specific failure mode: memories are anchored to git blob hashes, so when the code they're about changes, the memory is automatically flagged for review instead of being silently trusted or silently lost.
Key Differentiators
Git-aware staleness detection
Memories anchored to git blob hashes. Code changes → auto-flag needs_review.
Committed reverts → auto-heal.
Evidence, not vibes
Memories carry attached evidence (test output, commits) with mechanically computed strength.
provisional → active only with real evidence.
Zero dependencies, fully local
Pure stdlib + SQLite (FTS5). No embeddings, no API keys, no network calls.
.cortex/memory.db is just a file.
Real lifecycle with audit trail
7 statuses (provisional → active → superseded → archived, etc.), every change requires
--reason, permanently recorded.
Measured Impact
~12% fewer fresh tokens, ~15% lower API-equivalent cost — at identical fix quality (30/30 correct, zero disqualifications)
30-session benchmark: 3 real historical bugs from the repo, re-introduced and solved 10× each by Claude Haiku 4.5, with and without a robo-cortex memory pack. Medians of paired per-task deltas, scaling up to −26% tokens / −39% cost on the task hardest to locate unaided. A recorded memory pays for itself after ~1–6 reuse sessions (median ≈ 5). No wall-clock claim — time deltas were noise-dominated. Known limits: one subject model, one well-documented codebase, 3 task types. See benchmark-results.md and EVALUATION.md for full methodology and raw evidence.
Quick Start
pip install robo-cortex
cd /path/to/your/project
roco init
roco record --type lesson --scope repo \
--statement "Copy-to-clipboard fails on HTTP; use textarea fallback." \
--confidence high
roco retrieve --task "copy to clipboard"MCP Configuration
Works with Claude Desktop, Claude Code, Cursor, Codex CLI, and any MCP-compatible agent. Listed on the
official Model Context Protocol Registry as
io.github.robotel-limited/robo-cortex.
Requires the mcp extra: pip install robo-cortex[mcp].
{
"mcpServers": {
"robo-cortex": {
"command": "roco",
"args": ["mcp", "--repo", "/path/to/project"]
}
}
}Assets
- Why I Built a Git-Anchored Memory for AI Coding Agents — launch article (dev.to)
- README
- PyPI package page
- EVALUATION.md
- benchmark-results.md — full methodology and raw evidence
- Read the paper (Zenodo, PDF) — "robo-cortex: Git-Anchored, Evidence-Based Memory for AI Coding Agents"
About Robotel Limited UK
Robotel Limited UK ([email protected]) builds open-source tooling for AI coding agents. More at robotel.top.
Media Contact
Email: [email protected]
GitHub: github.com/robotel-limited/robo-cortex