AlekseiMarchenko/central-intelligence

v0.4.0 Security

This release includes 1 security fix for security teams reviewing exposed deployments.

Published 2mo AI Agents & Assistants

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

This release patches 1 known CVE

Affected surfaces

auth deps

Summary

AI summary

Fixed BM25 full-text search, removed insecure trigram search and context compression.

Full changelog

What's New

Retrieval actually works now. BM25 full-text search was broken since launch. Trigram search ran against encrypted ciphertext. Context compression silently forwarded decrypted memories to OpenAI. All three are fixed. Benchmark score: 92/100 → 100/100.

Fixed

BM25 full-text search — tsvectors now populated from plaintext before encryption
Removed trigram search (ran against ciphertext, producing noise)
Removed context compression (sent decrypted memories to OpenAI, breaking encryption guarantee)
Batch N+1 queries in recall with WHERE id = ANY(...)
Specific error handling instead of bare catch {} blocks
forget() now logs usage events
Migration uses CREATE INDEX CONCURRENTLY to avoid table locks
Lazy tsvector backfill during recall for old memories

Added

Vitest test suite (32 tests)
GIN index on content_tsv
VERSION file and CHANGELOG.md

Changed

Hybrid retrieval: vector + BM25 (was vector + BM25 + trigram)
All packages unified to v0.4.0

Breaking Changes

Removed trigram search (previously ran against encrypted ciphertext)
Removed context compression (previously forwarded decrypted memories to OpenAI)

Security Fixes

Removed context compression that sent decrypted memories to OpenAI, restoring encryption guarantee

View diff on GitHub

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

Share on X Share on Bluesky

Track AlekseiMarchenko/central-intelligence

Get notified when new releases ship.

About AlekseiMarchenko/central-intelligence

Persistent memory for AI agents. Five tools (remember, recall, context, forget, share) with semantic search via vector embeddings and agent/user/org scoping. Works with Claude Code, Cursor, Windsurf, and any MCP client.

All releases →