Core Team
The humans behind the contamination firewall.
J
Jayson
Project Lead / Architect
Designing memory systems that earn trust through proof. Building mem20 because agents deserve memory they can audit.
C
Coach
VP Bot Relations
Runs the multi-agent fleet on mem20 daily. Tests concurrency, validates cognitive features, documents what breaks.
Q
QAMaster
VP Quality
Owns the audit suite, contamination tests, and QA gates. Will not ship if contamination_rate != 0.
B
Blendie
VP Creativity
Leads creative and 3D integrations. Blender module, Unity module, and the design tools roadmap.
Contribute
mem20 is MIT-licensed and open to contributions. Here's how to get involved:
- Code: Fork the GitHub repo, pick up a TODO-tagged issue, and open a PR
- Docs: Improve the README, add examples, write guides — docs are code too
- Integrations: Use the blank integration template to add support for your favorite tool
- Beta testing: Sign up, run mem20 in your stack, report what breaks
- Research: Cognitive architecture, memory grounding, epistemic safety — we want to hear your ideas
Development Note
Engine logic lives in memory_engine/memory.py, not the MCP server. MCP tools are thin handlers. To contribute effectively, start by reading the store engine, not the server layer.
Adding a New Integration
- Create
mcp/tools/<name>_tools.py with a <Name>ToolsMixin
- Guard any external executable with a graceful check
- Add the mixin to
Mem20MCPServer's bases in server.py
- Call
self.register_<name>_tools() in _setup_tools()
- See
IntegrationToolsMixin.register_integration_template() for the scaffold