This release includes 1 security fix for security teams reviewing exposed deployments.
Topics
+2 more
Affected surfaces
Summary
AI summaryUpdates Highlights, Sonnet/Haiku/Opus, and GPT-4o/4o-mini/4.1 across a mixed release.
Full changelog
Highlights
- Provider Compatibility CI fully green — daily matrix across Anthropic (Sonnet/Haiku/Opus), OpenAI (GPT-4o/4o-mini/4.1), Google (6 Gemini catalog models incl. embeddings) and NVIDIA NIM; rotated CI credentials, de-flaked free-tier throttle handling (429/503 treated as transient), redacted failure logs.
- Security hardening — removed leaked env backup, hardened .gitignore, gitleaks scanning with curated allowlist on every push, registry port bound to localhost.
- 72 governed tools synced with the live control plane (mcp.thinkneo.ai) — all validated post-audit: injection patterns expanded, compare_models static catalog, SIEM export formats, enterprise billing handling, OAuth client-name sanitization, client-IP capture.
- Test suite: 299 unit tests green — async tool harness fixed, bridge/A2A test args aligned with real tool signatures.
- CI on Node 24 — actions/checkout v7, setup-python v6, upload-artifact v7 (#5); removed duplicate legacy registry publish workflow.
- SDK — npm package name corrected to @thinkneo_ai/sdk.
Registry version: ai.thinkneo/control-plane 1.29.0.
Note: v3.13.0 tag is a historical April benchmark tag without a release; this release supersedes it.
Security Fixes
- Removed leaked env backup, hardened .gitignore, added gitleaks scanning on every push, bound registry port to localhost
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 ThinkneoAI/mcp-server
ThinkNEO Control Plane — Enterprise AI governance MCP server with runtime guardrails, observability, AI FinOps, and agent lifecycle control.
Related context
Beta — feedback welcome: [email protected]