Chronic reminiscence for Claude Code, Codex & coding brokers | Agentmemory

Hello Product Hunt 👋

I constructed AgentMemory as a result of coding brokers nonetheless have one painful limitation: they overlook between classes.

• You give an explanation for your structure as soon as.

• You debug a manufacturing factor as soon as.

• You make a decision on a library or development as soon as.

Then the following consultation begins from 0 once more.

AgentMemory provides AI coding brokers power reminiscence throughout classes, so they are able to in truth construct on what they’ve already realized about your codebase. It mechanically captures what your agent does, compresses it into structured reminiscences, indexes them with hybrid seek, and injects the fitting context again into long run classes.

It really works with Claude Code, Cursor, Codex CLI, Gemini CLI, Windsurf, Kilo Code, OpenCode, Cline, Roo, Goose, Aider, Hermes, OpenClaw, and principally any MCP or REST-capable agent.

From day one, I sought after it to be:

• 100% open supply

• Unfastened to run in the neighborhood

• No exterior database required

• Works by way of MCP, REST, and easy hooks

• Constructed for actual coding workflows, now not toy “chat historical past” reminiscence

On benchmarks, AgentMemory will get 95.2% R@5 and 98.6% R@10 at the LongMemEval-S retrieval suite the usage of BM25 + vector seek, whilst reducing context utilization by way of round 92%.

Fast get started:

Run: npx @agentmemory/agentmemory

In the event you are living to your coding brokers on a daily basis, that is for the instant you assume: “Wait, I already defined this the day prior to this.”

Would like comments from developers, heavy agent customers, and open‑supply maintainers.

GitHub: https://github.com/rohitg00/agentmemory