MuscleMemory is open-source procedural memory for AI agents that learns from what worked and what failed. Coding agents relearn the same things every session, so MuscleMemory watches your agent work, and when a task really succeeds it learns how the task was done. The next time a similar task comes up it hands the agent a short note. It is local by default and Apache-2.0 licensed.
Be the first to rate MuscleMemory. Your feedback helps other users decide.
When the tests pass after the last change, MuscleMemory records how the task was done.
A similar task later gets a short note handed to the agent.
Local setup, Docker and PostgreSQL are all documented, with MCP and SDK integrations.
Similar tools in the same category — compare side-by-side
ReliaMesh is open-source reliability infrastructure for AI agents. It ingests structured outcomes, d...
CommonForge is for many agents and people changing one project at once, running on Cloudflare Worker...
Hatoba is an open-source desktop SSH client with an AI assistant and end-to-end encrypted sync throu...
Guides and comparisons about code tools
Ten Show HN launches that answer the same question differently: where does what an agent learned live, who owns it, and who can read it back?
Nine tools for supervising coding agents: phone alerts when an agent blocks, review inboxes, menu-bar contacts, shared scratchpads and team boards.
Nine tools that constrain what an agent can produce and reach: slop-detecting lint rules, declared sandbox tiers, cheaper crawling and signed messaging.
Twenty-five coding tools sorted by what they do: autocomplete, codebase comprehension, testing, or self-hostable models, from CodeGeeX to Codestral.
Coding agents increased output but not shipped software. The gap is verification, context and drift. Here is what converts agent output into working systems.
