Mnemo
Autonomous bitemporal memory engine for AI coding agents.
Mnemo provides persistent, cross-session architectural memory. It automatically maps repository structure, tracks context across edits, and exposes structured retrieval via Model Context Protocol (FastMCP) and clean CLI tools.
Key Features
- Autonomous AST Mapping: Automatically parses modules, classes, and dependencies with incremental SHA-256 diff caching.
- Bitemporal Knowledge Engine: Tracks both event time and assertion time using SQLite with WAL mode and full-text search (FTS5).
- Self-Healing Storage: Dynamic schema migrations and automatic FTS index repair.
- Dark Minimal Visualizer: Interactive 3-column Sankey graph mapping categories, memory tiers, and architecture flows.
Installation
pip install mnemo
Quick Start
1. Initialize Memory Store
mnemo init
2. Autonomous Repository Scan
mnemo scan .
3. Interactive Knowledge Visualization
mnemo visualize
4. Health Diagnostics
mnemo doctor
Model Context Protocol (FastMCP)
Mnemo integrates natively with Claude Desktop, Cursor, Antigravity, and any MCP client over stdio:
{
"mcpServers": {
"mnemo": {
"command": "mnemo",
"args": ["serve"]
}
}
}
Available MCP Tools
mnemo_remember: Stores or updates facts through the AUDN classifier and auto-links mentioned code entities.mnemo_search: Hybrid 3-channel retrieval (Vector + FTS5 + Recursive CTE Graph) fused via Reciprocal Rank Fusion (RRF).mnemo_scan_project: Triggers an on-demand incremental AST scan of the repository.mnemo_invalidate: Soft-deletes records with bitemporal invalidation timestamps.mnemo_tier_decay: Recalculates Ebbinghaus salience decay and tier promotions/demotions.mnemo_get_debt_ledger: Detects decaying working-tier facts, stale core assumptions, and architectural debt.
Architecture
- Storage: Local SQLite in WAL mode with FTS5 BM25, vector embeddings, and recursive CTE graph traversal.
- Serialization: Token-Optimized Object Notation (TOON) providing up to 56% token savings over standard JSON.
- CLI: Borderless Matrix interface powered by Typer and Rich.
License
MIT License. Copyright (c) 2026 Mnemo Contributors.
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file mnemo_mcp_agent-0.3.2.tar.gz.
File metadata
- Download URL: mnemo_mcp_agent-0.3.2.tar.gz
- Upload date:
- Size: 64.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/7.0.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
1ab4aada0698da79a2e0d426ff0e4409aa7fa47832b2a69a485d68ffdf2d5a2f
|
|
| MD5 |
e812ce0c847bbc46bf1f7df459b1baa9
|
|
| BLAKE2b-256 |
ddd02b793dab8937f3b2ddd886afcfd6d68e7f615e0fb89aedb00fbc9eaed690
|
File details
Details for the file mnemo_mcp_agent-0.3.2-py3-none-any.whl.
File metadata
- Download URL: mnemo_mcp_agent-0.3.2-py3-none-any.whl
- Upload date:
- Size: 52.9 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/7.0.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
ef9ec71ae372fd73e8696475a5a0c95a7bd2c42790cf9a139e734be760beac57
|
|
| MD5 |
0e4e90f6515fba36a9287ff4b39670e1
|
|
| BLAKE2b-256 |
e90d4838d9d683f7a762e916a5d74c88e529550fc905559a12696071a9a19186
|