Skip to content

feat(registry): Check for changes scans the pinned card's skills again - #1355

Merged
vgmartinez merged 1 commit into
developfrom
feat/a2a-plane-fumm19
Nov 14, 2026
Merged

vgmartinez merged 1 commit into
developfrom
feat/a2a-plane-fumm19

Conversation

@vgmartinez

Copy link
Copy Markdown
Contributor

Summary

Agent skills were scored by TrustGuard only when a card was previewed, pinned or approved, so an agent pinned before the scan existed never got verdicts, and a pinned card could not be checked again over time.

  • POST …/registries/{id}/agent-card/refresh now scans the pinned card's skills on every call and saves the verdicts with it (PinA2ACard from the same digest, keeping card_pinned_at; nothing is saved if the card was pinned again meanwhile). Registry cache invalidated as approve does.
  • The response adds skill_scan (the pinned card's new verdicts). fetched_skill_scan is unchanged: the live card's skills, when it changed.
  • A scan that cannot be scanned or saved does not fail the refresh: it is logged (agent skill scan not saved) and the comparison is answered.
  • NewRefreshAgentCard takes the pinner, cache manager, publisher and logger, as NewApproveAgentCard does.
  • Docs: docs/a2a.md (Card pinning, Skill scan).

Console: NeuralTrust/app (Skills tab shows the new verdicts after Check for changes).

Testing

  • go test ./pkg/app/registry/... ./pkg/container/... ./pkg/api/handler/http/registry/...
  • New: refresh rescans the pinned card and saves the verdicts with the same pin time; refresh answers when the save is refused.

Generated by Claude Code

Skills were scored only when a card was previewed, pinned or approved. A
refresh now scans the pinned card's skills again and saves the verdicts with
it, unless the card was pinned again meanwhile, so an agent's skills can be
checked over time and agents pinned before the scan existed get verdicts. A
scan that cannot be saved is still answered.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KF89yEyaPDgX5rDdR7NMt9
@github-actions

Copy link
Copy Markdown
Contributor

Trivy code scan — HIGH/CRITICAL

No HIGH or CRITICAL vulnerabilities found. (run)

@vgmartinez
vgmartinez merged commit 45e8c07 into develop Nov 14, 2026
14 of 16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants