Developer OS is the control plane for an autonomous AI engineering team — identity, plugins, knowledge, and tools assembled into one persistent context, with every verdict traceable to evidence.
❯ modules
The instrumentation an autonomous AI engineering team runs on — each module wired into the same persistent context.
Persona, occupation, and core instructions assembled into one persistent context the model loads on every session.
Hybrid scoring — IDF-weighted keywords blended with pgvector similarity — ranks the right plugins for the task at hand.
Domains of persistent, embedded memory. Recall blends keyword and semantic search so context survives across sessions.
Tools exposed over the Model Context Protocol across infrastructure, AI, and search servers — connected, authed, ready.
RLS policies, SECURITY DEFINER RPCs, rate limiting, and dual auth. The control plane fails closed, by design.
Token budgets, usage patterns, and the evidence-driven Verification Engine — every quality verdict is traceable.
❯ pipeline
30 tools exposed over the Model Context Protocol. Runs on your VPS behind dual auth.
A 7-component scorer blends IDF keywords, vector similarity, categories, and dependencies into one ranked context.
Pattern detection across sessions. The control plane measures its own verdicts and tightens over time.