Repository navigation
fix: show cloud workspace test results on Heart board #265
Description
Activity
The published Heart board says the workspace test-run is unobserved even when its cloud workflow ingested a measured report. The renderer now treats
test_runas cloud-observed, displaying the measured 1,405-pass result; if no report exists, it shows an honest unknown card. Existing failure and conclusion-only count handling remains in place.Closes #265.
API Changes
No library API changes. Only the Heart board presentation and focused tests change.
Validation
- Dashboard suite: 128 passed, including measured cloud success, measured failure, missing evidence and conclusion-only counts.
- Full Heart suite and tenant-firewall checks run on the branch; see issue validation comment for the final count.
git diff --checkpassed. Local Heart readiness was GREEN/100 after a fresh tick.
Publication
After merge, refresh the local observations through
pyauto-heart tickand the privacy-scrubbedpublishpath, then runheart-health.ymland verifyboard.json. This PR makes no changes to thresholds, weights or check outcomes. Existing dirty worktrees are preserved.The published Heart board says the workspace test-run is unobserved even when its cloud workflow ingested a measured report. The renderer now treats
test_runas cloud-observed, displaying the measured 1,405-pass result; if no report exists, it shows an honest unknown card. Existing failure and conclusion-only count handling remains in place.Closes #265.
API Changes
No library API changes. Only the Heart board presentation and focused tests change.
Validation
- Dashboard suite: 128 passed, including measured cloud success, measured failure, missing evidence and conclusion-only counts.
- Full Heart suite: 1,143 passed. Tenant-firewall check passed.
git diff --checkpassed. Local Heart readiness was GREEN/100 after a fresh tick.
Publication
After merge, refresh the local observations through
pyauto-heart tickand the privacy-scrubbedpublishpath, then runheart-health.ymland verifyboard.json. This PR makes no changes to thresholds, weights or check outcomes. Existing dirty worktrees are preserved.Cloud test-run card reflects observed smoke results
PyAutoHeart PR #266 makes the published board render measured workspace test results already ingested by the cloud health workflow. Missing artifacts remain unknown; measured failures and conclusion-only counts keep their existing semantics. No release score, threshold, skip or weight changed.
Validation: 1,143 Heart tests passed in an isolated worktree with the required Brain sibling; dashboard focused tests 128 passed; tenant firewall and diff-check passed. The previous shell path-test failure came from the missing sibling in the first isolated run, and passed unchanged once the normal worktree layout was restored. User authorized end-to-end work including CI judgement, merge, and dashboard publication.
The older dev-box worktree/timing/profiling observations are advisory and will be refreshed with the privacy-scrubbed
pyauto-heart tick && pyauto-heart publishafter merge. User worktrees remain intact. The task worktrees have only reproducible Python/pytest caches; evidence and logs are stored outside them.
Overview
The published Heart board masks a measured cloud workspace-smoke result as “unobserved.” The daily workflow already ingests its artifact; the renderer still classifies
test_runas local-only.Plan
test_runsection using existing readiness/count logic.Detailed implementation plan
Primary repo: PyAutoHeart. Branch: feature/cloud-test-run-card. Canonical main clean at the branch survey; no active Heart claim. Change
heart/dashboard.pyonly for classification/fallback. Updatetests/test_dashboard.pywith the measured cloud and missing evidence cases. Run focused and Heart tests plus tenant firewall. Preserve all thresholds and scores. After merge runpyauto-heart tick, reviewpublish --dry-run, publish local-only sections, and verify liveboard.json.Original Prompt
Approved user request
Show cloud-observed workspace tests on the Heart board
Type: bug
Difficulty: small
Autonomy: human-required
Original request
Evidence and scope
@PyAutoHeart published board 2026-10-01T19:55:48Z is GREEN/100 but the Test run section is unobserved. The same workflow log says
test_run ready 1405p/0f/122s @ cloud#36909756099.heart/dashboard.pyleavestest_runinLOCAL_ONLY_FAMILIES, which unconditionally masks the measured snapshot. The dev-box observation was last published 3d ago; worktree drift, script timing and profiling drift remain local-only. Refresh those through the existingpyauto-heart tickandpublishdoor after merge, preserving privacy scrub and user worktrees.Plan approved in chat
The user explicitly requested end-to-end completion without asking. For the display fix: remove test_run from the local-only list, let the existing measured renderer surface its ready/failure/unknown state, and keep an unobserved placeholder when no test-run evidence exists. Update focused board tests for measured and missing evidence, including nonfabricated counts. Run Heart suite and tenant firewall, open PR, judge every CI leg, merge if green, close issue and Mind task, and publish/check the live board. No score, threshold, skip or weight changes. Older dev-box sections remain advisory with age shown and are refreshed only through sanctioned publish.
Branch survey
Heart canonical clean main after fast-forward; no other Heart claim. Mind canonical branch belongs to another session; isolated ledger is codex/cloud-test-run-card. Source branch feature/cloud-test-run-card from origin/main.