HiveMind MCP
Persistent swarm memory through Obsidian — powered by your preferred LLM providers, gated by a GitHub Sponsors subscription.
Created by HaChazal (Chaz Leland Hamm) — Metanoia Unlimited LLC. This is the memory layer of the LeviathanTalon swarm. Every agent, every conversation, every decision lives in your Obsidian vault and is recalled semantically — across any provider you already pay for.
Why this exists
AI agents forget. HiveMind MCP makes them remember — permanently, locally, in your own Obsidian vault, with semantic recall across all your preferred LLM providers (OpenAI, Anthropic, Gemini, Grok, DeepSeek, Mistral, Cohere, Cerebras, Together, OpenRouter, Braintrust) routed through the Leviathan Key Rotator so you never burn a single key dry.
What it does
- Persistent memory — reads/writes
HiveMind.mdand the full vault (journals, entities, decisions) through Obsidian. - Semantic recall — embeddings + cosine similarity, not just keyword grep. Ask "what did we decide about the video pipeline?" and get the actual decision, even if the words don't match.
- Provider-agnostic — one interface, your preferred providers, key-rotated and budget-capped.
- Subscription-gated — a GitHub Sponsors subscription unlocks the paid provider tier. Free tier works fully local (Ollama) with no API costs.
Credit
This project is the work of HaChazal (Chaz Leland Hamm), Principal Operator of the LeviathanTalon swarm, founder of Metanoia Unlimited LLC. The HiveMind vault architecture, the key rotator, the swarm memory protocol, and the Obsidian integration are all his design. If you use this, star the repo and follow HaChazal — the swarm runs on community.
Install
pip install leviathan-hivemind-mcp
The distribution is leviathan-hivemind-mcp on PyPI; the command stays hivemind-mcp. Zero-config launch from anywhere (auto-downloads on first use):
uvx --from leviathan-hivemind-mcp hivemind-mcp
Or clone and run from source:
git clone https://github.com/chazam41892-gif/leviathan-hivemind-mcp
cd leviathan-hivemind-mcp
pip install -e .
Configure
Set your vault and provider keys — or drop them into the unified
~/.leviathan/HiveMind/.obsidian/HiveMind.env, which every HiveMind MCP
server loads automatically (front-end ↔ back-end):
export HIVEMIND_VAULT=~/.leviathan/HiveMind # your Obsidian vault
export HIVEMIND_OWNER=1 # owner mode: skip subscription gate
export LEVIATHAN_KEY_OPENAI_1=sk-... # key rotator format
export LEVIATHAN_KEY_ANTHROPIC_1=sk-ant-...
export LEVIATHAN_BUDGET_OPENAI=10.0 # optional budget caps
Run
hivemind-mcp
Register it in your MCP client (Claude Desktop, Cursor, OpenCode, etc.):
{
"mcpServers": {
"hivemind": {
"command": "hivemind-mcp",
"args": []
}
}
}
Tools
| Tool | Description |
|---|---|
read_shared_memory |
Read HiveMind.md shared swarm state |
read_soul |
Read SOUL.md — swarm identity + doctrine |
append_to_journal |
Append a memory to an agent's journal (append-only) |
recall |
Keyword or semantic recall across the vault |
semantic_recall |
Embedding-based recall (needs a provider or local model) |
read_entity / upsert_entity |
Read / layer canonical entities |
read_agent_identity |
Read an agent's identity.md |
read_openai_hivemind |
Read the OpenAI/Codex shared operating memory |
record_openai_highlight |
Write a resumable project capsule → 4 replicas |
chat |
Ask any preferred provider (gated by subscription) |
list_providers |
Show configured providers + key health |
License
MIT — but the credit is HaChazal's. Star the repo. Follow the swarm.
Built by HaChazal — Metanoia Unlimited. The swarm remembers.
Metadata
Release files for leviathan-hivemind-mcp 0.2.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| leviathan_hivemind_mcp-0.2.1.tar.gz | 21.8 kB | Without provenance | Details for leviathan_hivemind_mcp-0.2.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| leviathan_hivemind_mcp-0.2.1-py3-none-any.whl | Python 3 | none | any | Details for leviathan_hivemind_mcp-0.2.1-py3-none-any.whl |
Total release size: 45.8 kB
Release files / leviathan_hivemind_mcp-0.2.1.tar.gz
| Download URL | leviathan_hivemind_mcp-0.2.1.tar.gz |
|---|---|
| Size | 21.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
903f4cf4c2ad6285eaae6feea5cbd0e19268243485ffd484cade692fcb0506cd
|
|
BLAKE2b-256 checksum How to use checksums |
92ba0f1e8da61244c895e06f87d5f396bbbca9e1e392d7d702035905d37159bb
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.7
|
Release files / leviathan_hivemind_mcp-0.2.1-py3-none-any.whl
| Download URL | leviathan_hivemind_mcp-0.2.1-py3-none-any.whl |
|---|---|
| Size | 24.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
bdf59fd80c95ad4bcfe399296d21490935d195bd3984ea24deb72a681b9e92b9
|
|
BLAKE2b-256 checksum How to use checksums |
bc11bf29a26e61bcf1ac070d3c8fcc9f9278761f6ac9304fb8e8d87ece973eb2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.7
|
Security for leviathan-hivemind-mcp 0.2.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.