This release includes 3 breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
+14 more
Affected surfaces
Summary
AI summaryRemoved sentence-transformers, numpy, embedding model code, /api/models, fetch-model CLI, semantic config keys, and dead cache files.
Full changelog
Removed
sentence-transformersandnumpydependencies; embedding model code,/api/models,fetch-modelCLI, and semantic config keys.- Dead embedding cache files (
embeddings.npy,embeddings.hash) from store init, gitignore, and storage listings. semantic_scopefrom write/consolidate APIs and MCP handlers; consolidate is section-scoped exact duplicate cleanup only.
Changed
- Install and CI are much lighter without PyTorch or embedding model downloads; search remains keyword/token matching with agent-side relevance.
- GitHub Actions uses
uvwith dependency caching, Python 3.12 only on PRs, path filters to skip docs-only runs, and Buildx cache for Docker smoke tests. - Documentation aligned with keyword-only runtime: README, COMPATIBILITY, SECURITY, and CONTRIBUTING.
Breaking Changes
- Removed `sentence-transformers` and `numpy` dependencies along with embedding model code, `/api/models`, `fetch-model` CLI, and all semantic config keys.
- Removed `semantic_scope` from write/consolidate APIs and MCP handlers; consolidate now performs only section-scoped exact duplicate cleanup.
- Dropped dead embedding cache files (`embeddings.npy`, `embeddings.hash`) from store initialization, gitignore entries, and storage listings.
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 LuizEduPP/rememb
Persistent memory for AI agents with sectioned entries (project/user/context/etc), semantic search, CLI, and per-project scope. Local JSON, zero config, no server required.
Related context
Related tools
Beta — feedback welcome: [email protected]