This release adds 2 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+14 more
Summary
AI summarydeep_scan LLM temperature set to 0 for stable findings and prompt enforces precision over recall
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Low |
deep_scan still detects real semantic vulnerabilities like IDOR and auth bypass deep_scan still detects real semantic vulnerabilities like IDOR and auth bypass Source: granite4.1:30b@2026-06-06-audit Confidence: low |
— |
| Bugfix | Medium |
deep_scan runs LLM at temperature 0 for stable findings deep_scan runs LLM at temperature 0 for stable findings Source: llm_adapter@2026-06-06 Confidence: low |
— |
| Bugfix | Medium |
deep_scan prompt prioritizes precision over recall, eliminating speculative flags deep_scan prompt prioritizes precision over recall, eliminating speculative flags Source: llm_adapter@2026-06-06 Confidence: low |
— |
Full changelog
- deep_scan now runs the LLM at temperature 0 — same code yields stable findings instead of varying run-to-run
- deep_scan prompt enforces precision over recall: only flags vulnerabilities present in the shown code, no speculation about unseen helpers, no generic hardening noise (clean code that drew 3 speculative findings now returns 0)
- Real semantic vulns (IDOR, business-logic, race conditions, auth bypass) still caught; validated live against the API
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 goklab/guardvibe
Security MCP for vibe coding with 330 rules and 29 tools. Purpose-built for AI-generated code — scans Next.js, Supabase, Clerk, Stripe, Prisma, Hono, GraphQL, and 25+ modules. Cross-file taint analysis, host security audit, auto-fix, SARIF export, pre-commit hook, and CVE version detection. Zero config, runs locally.
Related context
Related tools
Beta — feedback welcome: [email protected]