Most MCP servers give AI tools.
OpenTology gives AI understanding — persistent memory, dependency awareness, and impact analysis across every session.
Not just tools — a fundamentally different way for AI to understand your project.
Decisions, issues, and knowledge persist across sessions in an RDF graph. Your AI remembers what it learned yesterday.
Before editing, AI checks the dependency graph to understand blast radius. No more accidental breakage of dependent modules.
Symbol-level scanning for TypeScript, Python, Go, Rust, Java, and Swift. Classes, functions, methods, and call graphs.
Oxigraph runs as WebAssembly. No containers, no native builds. Works everywhere Node.js runs.
Automatically injects CLAUDE.md so AI checks impact, searches context, and records decisions without being asked.
Built on RDF, SPARQL, RDFS reasoning, and SHACL validation. Query your graph with any SPARQL-compatible tool.
Real experiment on this codebase. Watch both approaches run.
"What breaks if I change store-adapter.ts?"
Three commands. That's it.
Point your AI assistant to OpenTology.
"command": "npx",
"args": ["-y", "opentology", "mcp"]
Creates graph, hooks, and AI instructions.
opentology context init
Builds the knowledge graph from your code.
opentology context scan