my-project-public

# Clawd Memory Documentation

Clawd Memory is the local brain for Clawd agents. It combines a Clawd-facing agent memory contract, a markdown vault, and a SQLite-backed memory engine.

#4

#5

Start here:

| Guide | Purpose |

|---|---|

| [Getting Started](getting-started.md) | Initialize the Clawd bank, write memories, recall context, and archive research |

#10

| [Architecture](architecture.md) | Understand the Clawd layers, vault, BEAM engine, and recall pipeline |

#11

| [Configuration](configuration.md) | Environment variables, vault paths, LLM consolidation, and Hermes config |

#12

| [API Reference](api-reference.md) | `ClawdBrain`, `clawd-brain`, and lower-level engine APIs |

#13

| [Hermes Integration](hermes-integration.md) | Use Clawd Memory through Hermes-compatible tools |

#14

| [LLM Installation Guide](llm-installation-guide.md) | Agent-oriented install and verification checklist |

#15

| [BEAM Benchmark](beam-benchmark.md) | Benchmark notes for the underlying memory engine |

#16

| [Comparison](comparison.md) | Trade-offs versus external/self-hosted memory providers |

Compatibility names:

| Name | Meaning |

|---|---|

| `clawd-brain` | Clawd-facing CLI |

#23

| `ClawdBrain` | Clawd-facing Python API |

#24

| `mnemosyne` | Underlying Python storage engine |

#25

| `mnemosyne_*` | Existing Hermes/MCP tool names kept for compatibility |

#26

#27

Use Clawd names in user-facing docs and agent workflows. Use Mnemosyne names only where the code, package, or integration contract requires them.

#28

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public