Scanned merlvintan/damn-vulnerable-llm-agent (LangChain-based DVLA fork): 2 findings, both HIGH-severity SQL injection via LLM-generated queries, governance score 27/100, EU AI Act readiness PARTIAL.

First attempt against szybnev/DVLA failed (no_agent_code — repo has no LLM/agent code); logged to data/scan-corpus-failures.json.

Corpus grows from 2 → 3 scans; total_findings_observed 9 → 11; average_governance_score 16.0 → 19.67.

Generated by Claude Code

Summary by cubic

Refreshes the scan corpus by adding a new scan for merlvintan/damn-vulnerable-llm-agent and introducing a failure log for rejected targets. Totals now 3 scans and 11 findings.

• New Features
  • Added scan for merlvintan/damn-vulnerable-llm-agent: 2 HIGH SQL injection findings; governance 27/100; EU AI Act readiness PARTIAL.
  • Created data/scan-corpus-failures.json logging a failed attempt for szybnev/DVLA (HTTP 400: no_agent_code) with fallback to merlvintan/damn-vulnerable-llm-agent.
  • Updated data/scan-corpus.json aggregates: total_scans 3, total_findings_observed 11, average_governance_score 19.67, last_refreshed 2026-06-28.

^{Written for commit c855d45. Summary will update on new commits.}