Cortex
Run a local-first persistent knowledge base with OWL-RL reasoning and a large MCP tool surface for agent-driven research and decisions.
Overview
Cortex is an MCP server for the Build phase that provides local-first persistent knowledge with OWL-RL reasoning across 22 MCP tools.
What is this MCP server?
- 22 MCP tools for knowledge capture, query, and OWL-RL reasoning
- Local-first persistent knowledge system (PyPI package abbacus-cortex v0.2.2)
- OWL-RL reasoning layer over stored concepts and relationships
- stdio MCP transport suitable for offline or privacy-sensitive workflows
- Open-source Cortex repository on GitHub for self-hosted deployment
- 22 MCP tools documented in server description
- PyPI package version 0.2.2 (identifier abbacus-cortex)
Community signal: 15 GitHub stars.
What problem does it solve?
Flat chat context cannot represent structured relationships or inference rules, so agents give shallow answers on complex domains.
Who is it for?
Technical solo founders building research-heavy or compliance-aware agents who want local ontology reasoning via MCP.
Skip if: Builders who only need quick markdown notes, pure embedding search, or fully managed cloud knowledge with zero local Python ops.
What do I get? / Deliverables
Your agent works against a reasoned knowledge graph on your machine with a broad MCP toolset for query and updates.
- Local Cortex MCP server with OWL-RL-backed knowledge store
- Agent-accessible graph query and update via 22 registered tools
- Repeatable structured memory layer for research and build decisions
Recommended MCP Servers
Journey fit
Structured knowledge and reasoning belong in Build when you instrument agents to cite facts and infer over ontologies, not only at launch. Agent-tooling covers MCP servers that give models structured memory, inference, and tool APIs on the dev machine.
How it compares
Local ontology and reasoning MCP with 22 tools, not a remote human-verification or trading arena server.
Common Questions / FAQ
Who is Cortex for?
Developers and indie AI builders who want local-first knowledge graphs and OWL-RL inference callable from coding agents.
When should I use Cortex?
Use it when agent tasks need structured entities, relationships, and rule-based reasoning rather than one-shot retrieval.
How do I add Cortex to my agent?
Install PyPI package abbacus-cortex, configure stdio MCP in your client per the Cortex GitHub repo, and enable the tool surface you need.