Fix cache severity bug, AI exit code, HTML crash, and score weighting gaps (2026.7.2)

[advaitpatel]

Summary

Six fixes found during manual CLI verification of the Phase 0-3 feature set, bundled as patch release 2026.7.2:

• ScanResultsCache keyed results by image name only, so scanning an image at a narrow --severity and then re-scanning at a wider severity silently served the stale narrower cache, dropping real HIGH/MEDIUM findings from the report. The cache key now includes the normalized severity list.
• AI analysis failures (bad provider/API key, model error) printed an error but still exited 0, contradicting the documented exit-code contract. AI failures now exit 3, same as scan failures.
• Trivy can return "Title": null for a finding; the HTML report writer called len() on it unconditionally and crashed generation for the whole report. Vulnerability ID, package name, installed version, and title are now null-safe in the HTML report.
• --compose --scan-only printed an unrelated "No image provided ... Running Dockerfile analysis only" message; now scoped to non-compose runs.
• When every per-service image scan in a compose file failed, the score calculator excluded the vulnerabilities axis from the weighted average even though compose static-misconfiguration findings existed — a compose file with multiple CRITICAL findings could score "GOOD". The vulnerabilities axis now always counts when findings exist.
• A Dockerfile with hardcoded credentials and other severe misconfigurations could still land in the mid-40s ("POOR" but not alarming) due to the blended score diluting the credential penalty. Overall score is now capped at 20/100 when a credential-looking ENV var is detected.

Version bumped to 2026.7.2 in setup.py; docs/CHANGELOG.md updated with a dated [2026.7.2] section.

Test plan

• pytest tests/ — 157 passed
• Reproduced and verified the cache/severity fix: scan itombot:1.0.5 at --severity CRITICAL, then --severity CRITICAL,HIGH,MEDIUM — second run no longer reuses the narrow cache and reports the full finding count
• Verified AI failure with no API key exits 3 (was 0)
• Verified HTML report generates without crashing on a python:3.9 scan containing a null-Title finding
• Verified --compose --scan-only no longer prints the Dockerfile message
• Verified compose scoring: insecure compose fixture with failed per-service image scans now scores 53/FAIR (was 80/GOOD)
• Verified credential-exposure cap: vulnerable-dockerfile fixture now scores 20/POOR (was 45/POOR); a clean fixture with no credentials is unaffected (100/EXCELLENT)

[github-actions]

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

OpenSSF Scorecard

Package	Version	Score	Details
actions/actions/first-interaction	753c925c8d1ac6fede23781875376600628d9b5d	🟢 4.6	Details Check Score Reason Maintained ⚠️ 0 0 commit(s) and 0 issue activity found in the last 90 days -- score normalized to 0 Dangerous-Workflow 🟢 10 no dangerous workflow patterns detected Packaging ⚠️ -1 packaging workflow not detected Code-Review 🟢 3 Found 1/3 approved changesets -- score normalized to 3 Binary-Artifacts 🟢 10 no binaries found in the repo CII-Best-Practices ⚠️ 0 no effort to earn an OpenSSF best practices badge detected Token-Permissions ⚠️ 0 detected GitHub workflow tokens with excessive permissions Pinned-Dependencies 🟢 3 dependency not pinned by hash detected -- score normalized to 3 License 🟢 10 license file detected Fuzzing ⚠️ 0 project is not fuzzed Signed-Releases ⚠️ -1 no releases found Security-Policy 🟢 9 security policy file detected Branch-Protection ⚠️ 1 branch protection is not maximal on development and all release branches SAST 🟢 9 SAST tool detected but not run on all commits

Scanned Files

• .github/workflows/welcome.yml