This release adds 1 notable feature for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Summary
AI summaryDefault embedding model changed to all-MiniLM-L12-v2 for improved retrieval quality.
Full changelog
What's new in v3.7.0
Bug fixes
CLAUDE_DESKTOP_REGISTERED: unbound variablecrash — parallel subshells cannot write back to the parent shell. All three registration flags are now initialised tofalsebefore the parallel phase launch.- 120s health timeout too short — on first install the embedding model is downloaded at container startup. Timeout extended to 5 minutes. A progress dot is printed every 10s so the terminal doesn't look frozen.
- Password/URL not shown — credentials are now fetched with a late retry after the parallel phases complete, and shown in the summary even if the health check timed out. Clear fallback commands shown if the container is still downloading.
New feature: embedding model selection
The installer now asks which model to use before starting the container:
Embedding model (downloaded on first start, cached on /data volume):
1) all-MiniLM-L12-v2 — fast, good quality [default]
2) all-MiniLM-L6-v2 — fastest, smaller (~80MB)
3) all-mpnet-base-v2 — best quality, slower (~420MB)
Default model changed
all-MiniLM-L6-v2 → all-MiniLM-L12-v2 — better retrieval quality, same speed class, only ~40MB larger.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About dl4rce/flaiwheel
Self-hosted memory and governance layer for AI coding agents. 28 MCP tools with structured knowledge capture, hybrid search (semantic + BM25 + cross-encoder reranking), behavioral documentation nudges, cold-start codebase analyzer, and git-native storage. Single Docker container, zero cloud dependencies.
Related context
Beta — feedback welcome: [email protected]