featured · github
claude-mem: Give Your Agents a Real Memory
Persistent context layer that compresses session data and reinjects it across runs. Cuts redundant context window waste and works across Claude, Gemini, Copilot, and more.
thedotmack/claude-mem ↗When you run an agent multiple times, it starts fresh each session. You end up re-explaining the same context, burning tokens, and the agent can't build on what it learned before.
Claude-mem captures everything your agent does during a session, compresses it intelligently, and stores it. Next run, relevant compressed context gets injected back in automatically. Same agent, zero amnesia.
This matters because: fewer wasted tokens means lower cost, faster responses, and agents that actually improve over repeated runs. Works with Claude Code, OpenAI, Gemini, Copilot, and others. If you're shipping multi-session workflows or long-running agent pipelines, this cuts friction and saves real money.
Share kit
Agents that remember: claude-mem adds persistent context across runs
Tired of re-explaining context every time your agent restarts. Claude-mem captures session data, compresses it, and reinjects relevant context into future runs. Token-efficient, multi-model support, and measurably cheaper per session. Worth testing if you're shipping agents that run more than once.
Your agent forgets everything after each session. Claude-mem captures what it did, compresses it, and injects it back next time. Same agent, no amnesia. Works with Claude, Gemini, Copilot, and more. Token savings add up fast. https://github.com/thedotmack/claude-mem
We shipped agents that ran multiple times. Every restart meant re-explaining context and burning tokens. Claude-mem solves this with a persistence layer that compresses session data and reinjects it intelligently. Real cost savings for multi-session workflows. Worth a look if you're scaling agents in production.
Just found claude-mem and it solves a real problem I've been sitting with: my agents forget everything between sessions. You run an agent, it learns context, you spin it up again tomorrow and start from zero. This captures what happened, compresses it with AI, then injects the relevant bits back in. Supports Claude, Gemini, Copilot, Hermes, and more. Means less token waste re-explaining the same context, fewer hallucinations from missing history, faster runs. If you're deploying agents in production, this cuts down the operational friction noticeably. https://github.com/thedotmack/claude-mem
agents that actually remember things between runs. claude-mem captures session context, compresses it, and injects it back when the agent spins up again. works with claude, gemini, copilot, hermes + more. less token waste. fewer gaps. feels simple but saves a lot of friction in production. https://github.com/thedotmack/claude-mem