This release includes breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
+13 more
Summary
AI summaryAdded the tech_stack_cve_audit composite tool with tier-aware behavior and cost enforcement.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Medium |
`tech_stack_cve_audit` tool added for one-call CVE audit across tech stack. `tech_stack_cve_audit` tool added for one-call CVE audit across tech stack. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Feature | Medium |
Product-name normalization invariant now covered by tests (case/alias). Product-name normalization invariant now covered by tests (case/alias). Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Feature | Medium |
'tech_stack_cve_audit' composite tool added for one‑call tech fingerprinting, CVE lookup, KEV cross‑ref, and exploit signals. 'tech_stack_cve_audit' composite tool added for one‑call tech fingerprinting, CVE lookup, KEV cross‑ref, and exploit signals. Source: granite4.1:30b@2026-05-22-audit Confidence: low |
— |
| Feature | Medium |
Tier‑aware behavior: Pro audits deeper set, lighter tier omits 'exploit_findings'. Tier‑aware behavior: Pro audits deeper set, lighter tier omits 'exploit_findings'. Source: granite4.1:30b@2026-05-22-audit Confidence: low |
— |
| Dependency | Medium |
No schema migration; no breaking changes introduced. No schema migration; no breaking changes introduced. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Performance | Medium |
Composite tool consumes exactly 10 credits, avoiding per-sub-call double charge. Composite tool consumes exactly 10 credits, avoiding per-sub-call double charge. Source: llm_adapter@2026-05-21 Confidence: low |
— |
| Bugfix | Medium |
Smithery probe fast-path preserved and verified to short-circuit before cost gate. Smithery probe fast-path preserved and verified to short-circuit before cost gate. Source: llm_adapter@2026-05-21 Confidence: high |
— |
| Refactor | Medium |
Drift-guard test ensures every composite tool has explicit cost mapping. Drift-guard test ensures every composite tool has explicit cost mapping. Source: llm_adapter@2026-05-21 Confidence: high |
— |
| Refactor | Low |
Wrapper calls internal implementation directly (no HTTP self‑hop). Wrapper calls internal implementation directly (no HTTP self‑hop). Source: granite4.1:30b@2026-05-22-audit Confidence: low |
— |
Full changelog
v1.33.0 — Tech-Stack → CVE Composite Tool
New MCP tool: tech_stack_cve_audit (MCP-only)
One-call composite that chains tech fingerprinting → product-name CVE lookup → KEV cross-ref → exploit signals for a target domain. Returns a unified cves_by_tech / kev_findings / verdict / summary report instead of forcing the agent to orchestrate 4+ separate tool calls.
- Cost: 10 credits, consumed exactly once (no per-sub-call double-charge).
- Tier-aware: Pro audits a deeper candidate set; the lighter tier runs a reduced candidate set and omits the
exploit_findingsfield entirely (absent from the wire via conditional serializer, not just nulled). - No HTTP self-hop: wrapper calls the internal impl directly (Pattern B).
Quality / hardening
- Drift-guard test enforces every composite tool is present in the cost map (CI fails if a new composite ships without an explicit cost).
- Product-name normalization invariant covered by tests (case/alias).
- Cherry-picked cleanly onto v1.32.7 — Smithery probe fast-path (
triggers/list/ai.smithery/events/list) preserved and verified to short-circuit before the cost gate (probes stay credit-free).
Counts
- MCP tools: 52 → 53 · Resources: 7 (unchanged) · Prompts: 3 (unchanged)
- Tests: 2404 → 2417 pytest (+13)
No schema migration. No breaking changes.
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 UPinar/contrastapi
Security intelligence API with 31 MCP tools for CVE/EPSS/KEV lookup, domain recon (DNS/WHOIS/SSL/subdomains/CT logs), IOC/threat intel, OSINT (email/phone/username), and code security scanning (secrets, injection). Free 100 req/hr.
Related context
Related tools
Earlier breaking changes
- v1.33.11 `bulk_sigma_rule_lookup` now costs 1 credit per `rule_id`, changing from flat 1 credit/call.
Beta — feedback welcome: [email protected]