docs: update ROADMAP — embed sidecar now In Progress, EPIC #117 with wave sub-issues #118 #119 #120

[four-bytes-robby]

Changes

• Mark Embedding Sidecar Architecture as 🔄 In Progress
• Link EPIC [EPIC] Go embed-sidecar: non-blocking real embeddings via llama.cpp HTTP server #117 and wave sub-issues [FEAT] Wave 1: Go embed-sidecar binary (status API, llama-server mgmt, model download, queue) #118 [FEAT] Wave 2: Plugin integration (build.ts, EmbeddingService sidecar mode, non-blocking embed, TUI) #119 [REFACTOR] Wave 3: Remove node-llama-cpp, modelDownloader.ts, cleanup env vars #120
• Document design decisions (two-mode, non-blocking, Go binary, port 8666, 30min idle timeout)

Closes #117

---

Summary by cubic

Update ROADMAP to mark Embedding Sidecar Architecture as In Progress and align it with EPIC #117 and waves #118–#120.

Document v1.7.1 decisions: Go sidecar managing llama-server; two modes (hash default, sidecar via BRAIN_EMBED_SIDECAR); non‑blocking embed(); prebuilt binary via releases; self‑bootstrapping downloads and caches model; port 8666 with single‑instance detection; 30‑min idle shutdown; rich /status with phase, progress, and queue depth.

^{Written for commit 6cd3c93. Summary will update on new commits.}

[cubic-dev-ai]

2 issues found across 1 file

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>