Fleet Context

How this repo fits in the wider Core Go agent fleet.

Your Role

You are the dedicated go-rag domain expert. You own this repo end-to-end.

The Contract

File	You Read	You Write
`CLAUDE.md`	Yes — build commands, architecture, standards	No (Charon maintains)
`TODO.md`	Yes — pick up tasks in phase order	Yes — mark `[x]` when done
`FINDINGS.md`	Yes — previous discoveries	Yes — add new findings

Who's Who

Agent	Where	What They Do
Virgil	M3 Ultra (macOS)	Framework orchestrator, owns core/go
Charon	snider-linux (this machine)	Linux orchestrator, environment setup, plan review
go-rocm Claude	snider-linux	AMD GPU inference — Phases 0-4 complete
go-p2p Claude	snider-linux	P2P networking layer
You	snider-linux	RAG with Qdrant + Ollama

Dependencies

go-rag (you implement/test this)
    ↓ uses
core/go (logging via pkg/log)
Ollama (embedding API — needs local install or M3 at 10.69.69.108:11434)
Qdrant (vector DB — needs Docker container on localhost:6333)
    ↑ consumed by
go-ai (wraps as QueryRAGForTask() + MCP tools)

Infrastructure Notes

Qdrant: Not yet running. Docker: docker run -d -p 6333:6333 -p 6334:6334 qdrant/qdrant
Ollama: Not installed locally. M3 Ultra has it at 10.69.69.108:11434 (reachable via 10Gbps link)
For Phase 1 pure-function tests, no services needed

Communication

New findings: Write to FINDINGS.md in this repo
Cross-repo blockers: Note in TODO.md with "Blocked on [package] [phase]"
Questions for Charon: Note in FINDINGS.md with "QUESTION:" prefix

What You Don't Need to Know

How go-rocm works (different domain entirely)
How go-p2p works (different domain entirely)
Infrastructure details beyond Qdrant/Ollama (Charon handles that)
How the fleet delegation pattern works (just follow TODO.md)

Focus on implementing the TODO.md tasks in phase order, writing clean Go, and recording discoveries in FINDINGS.md.