fix(app): show evidence block for failed automation runs

[tofikwest]

Summary

Follow-up to #2643. The automation-run detail card in Compliance → Task → Integration Checks rendered a View Evidence expandable JSON tree for every passing result but never for a failing one — even though the backend saves the same `evidence` payload in both cases and the API returns it identically.

After the Dependabot severity gating in #2643, fails surface exactly the context users need to understand them (the `open_by_severity` breakdown, `checked_at`, the `alerts` totals). Hiding that JSON for fails defeats the point of showing the evidence at all.

This PR mirrors the passing block's `

Details

> ` pattern onto the findings map so both states render identically. 12-line diff, one file.

Trace of the pre-existing inconsistency:

• Check emits `evidence` identically for pass/fail (`packages/integration-platform/src/manifests/github/checks/dependabot.ts`)
• Backend save path writes `evidence: JSON.parse(JSON.stringify(finding.evidence || {}))` for both (`apps/api/src/trigger/integration-platform/run-connection-checks.ts:191-214`)
• DB schema has single `evidence Json?` column shared by pass/fail (`IntegrationCheckResult`)
• API endpoint returns `evidence: r.evidence` with no pass/fail filtering (`task-integrations.controller.ts:614`)
• Frontend hook types `evidence?: Record<string, unknown>` on all results (`useIntegrationChecks.ts:54`)
• Frontend component only renders it in the passing map (`TaskIntegrationChecks.tsx:1077-1088`) — this was the only gap, now closed

Test plan

• `tsc --noEmit` clean for this file's module graph (unrelated pre-existing failures in `auth-client.ts` and `ai-elements/tool.tsx` already present on `main`)
• Diff is additive only; no existing behaviour changed for passing results, logs, remediation, or badges
• Before merge: run a Dependabot automation against a repo with open high alerts and visually confirm the red failing card now shows the same "▶ View Evidence" expandable block that passing cards show

Why not also write a unit test?

`TaskIntegrationChecks.tsx` is a 1432-line component with no existing test file of its own. Adding one purely for this 12-line conditional would be out-of-scope scaffolding; the sibling component tests (`SingleTask.test.tsx`, etc.) don't reach into this code path. Happy to add one if you'd prefer — flag it in review.

🤖 Generated with Claude Code

---

Summary by cubic

Show the "View Evidence" JSON block for failing automation runs in Compliance → Task → Integration Checks, not only for passing results. This fixes the UI inconsistency and lets users see the same evidence (e.g., open_by_severity, checked_at) to understand failures.

^{Written for commit 975b4c9. Summary will update on new commits.}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
app	Ready	Preview, Comment	Apr 23, 2026 3:04am
comp-framework-editor	Ready	Preview, Comment	Apr 23, 2026 3:04am

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
portal	Skipped		Apr 23, 2026 3:04am

[cubic-dev-ai]

No issues found across 1 file

_{Auto-approved: Low-risk UI fix that shows existing evidence data for failed checks, mirroring the behavior of passing checks. No business logic or data integrity changes.}

[claudfuen]

🎉 This PR is included in version 3.30.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀