# Honcho

> Honcho is an open-source memory layer for AI agents, built by Plastic Labs. It gives agents a persistent, reasoning-based memory of each user across sessions, tools and models. State of the art on LongMem, LoCoMo and BEAM. Self-hostable (AGPL-3.0) or managed at honcho.dev.

## Start here

- [Letta vs Honcho](https://honcho.dev/vs/letta): an agent that edits its own memory vs memory shared by the agents you run, pricing, self-hosting, benchmarks, with sources
- [Honcho](https://honcho.dev/): what Honcho is, plugins for Claude Code, Codex, OpenClaw, Hermes and DeepSeek Harness, pricing
- [Supermemory vs Honcho](https://honcho.dev/vs/supermemory): user profiles and hybrid search vs peers, four meters vs one, MIT SDKs and a closed server vs AGPL, benchmarks, with sources
- [Claude Code memory across every project](https://honcho.dev/claude-code-memory): a Claude Code plugin that carries how you work across repos and machines, versus CLAUDE.md and auto memory
- [One memory for every AI tool](https://honcho.dev/ai-memory): portable memory across Claude Code, Cursor and Codex
- [OpenViking vs Honcho](https://honcho.dev/vs/openviking): context filesystem vs memory of people, self-hosting, licenses, benchmarks, with sources
- [Honcho for Hermes Agent](https://honcho.dev/hermes): a memory plugin for Hermes, set up in one command
- [Honcho for the Pi coding agent](https://honcho.dev/pi): a memory extension for Pi, set up in one command
- [Honcho CLI](https://honcho.dev/cli): inspect, query and manage Honcho from the shell
- [Mem0 vs Honcho](https://honcho.dev/vs/mem0): how each builds memory, pricing, self-hosting, licenses and benchmarks, with sources and where Mem0 fits better
- [Benchmarks](https://honcho.dev/evals/): LongMem, LoCoMo and BEAM results
- [Hindsight vs Honcho](https://honcho.dev/vs/hindsight): memory banks vs peers, retain vs ingest pricing, MIT vs AGPL, benchmarks, with sources

## Documentation

- [Docs index (llms.txt)](https://honcho.dev/docs/llms.txt)
- [Overview](https://honcho.dev/docs/v3/documentation/introduction/overview.md)
- [Quickstart](https://honcho.dev/docs/v3/documentation/introduction/quickstart.md)
- [Claude Code integration](https://honcho.dev/docs/v3/guides/integrations/claude-code.md)
- [Migrating from Mem0](https://honcho.dev/docs/v3/guides/migrations/mem0.md)
- [Self-hosting](https://honcho.dev/docs/v3/contributing/self-hosting.md)

## Blog

- [Blog index (llms.txt)](https://honcho.dev/blog/llms.txt)
- [Benchmarking Honcho](https://honcho.dev/blog/blog/benchmarking-honcho)
- [Memory as Reasoning](https://honcho.dev/blog/blog/memory-as-reasoning)

## Code

- [GitHub](https://github.com/plastic-labs/honcho)
- [MCP server](https://honcho.dev/docs/v3/guides/integrations/mcp.md): connect any AI tool to Honcho via mcp.honcho.dev