This release fixes issues for SREs watching stability and regressions.
✓ No known CVEs patched in this version
Summary
AI summaryFixed MCP reconnect logic to survive client disconnects and added elapsed time logging for embedding model readiness.
Full changelog
🔄 MCP Reconnect Loop + Logging
Critical Fix: MCP Connection Stability
The previous reconnect logic was a stub that always exited — when the MCP client disconnected, the server would wait for a timeout and then shut down. Now, the server implements a proper reconnect loop:
- Client disconnects → server detects via
service.waiting() - Brief 500ms pause → re-creates
stdio()transport - Serves a new MCP session with the same state/data
- If first reconnect fails → waits
reconnect_timeoutseconds → retries once - Server only exits on SIGINT/SIGTERM or double reconnect failure
This ensures 100% MCP connection availability — the server survives client disconnects without losing any indexed data.
Improved Logging
Embedding model readynow includeselapsed_sec(e.g.,elapsed_sec="8.7") for visibility into model loading performance
Full Changelog: https://github.com/pomazanbohdan/memory-mcp-1file/compare/v0.2.7...v0.2.8
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 pomazanbohdan/memory-mcp-1file
A self-contained Memory server with single-binary architecture (embedded DB & models, no dependencies). Provides persistent semantic and graph-based memory for AI agents.
Related context
Beta — feedback welcome: [email protected]