Fix PMO UI white-screen on load + Windows-broken Playwright utils - #112
Merged
Conversation
… recursion, policy visibility (bd-6z9) - audit_coverage_requirement now matches pack:* presets risk-gated to HIGH/CRITICAL - single recursive sub_team collector shared by coverage and mismatch checks - headless audit_missing defect made actionable (outcome parity with interactive path) - policy validation failures logged + surfaced in score_warnings instead of swallowed - _IMPLEMENT_PHASE_KEYS derived from phase_roles.IMPLEMENT_PHASE_NAMES (adds draft) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…bd-5mq) - _upsert_plan: INSERT OR REPLACE -> ON CONFLICT(task_id) DO UPDATE excluding release_id - legacy JSON->SQLite migrator binds all plan columns instead of 10 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…trict-backend errors (bd-85c) - resume() saves execution state after the resolver loop, mirroring siblings - _audit_step_agents returns first EXISTING dir's audit, not first dir with problems - execute CLI maps UnknownTeamBackendError to a clean user error (no traceback) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
- active-task sqlite probe resolves BATON_DB_PATH first, matching bead_cmd precedence - planner_validation details carry plan_selection (active-task vs fallback-first-found) with a human-readable caveat when validating a fallback guess Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…up missing roots (bd-j9f) - per-pack OSError isolation mirroring the registry guard from 129bac4 - +.md stem fallback no longer satisfies declarations with known non-md extensions - repeated --knowledge-root paths yield a single missing-root issue Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…c40) Matches forge_plan/forge_regenerate: PlanQualityError -> 422 via plan_quality_error_detail before the generic 500 fallback. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…2hs) Mirrors next_actions()' diff-based conditional save: the post-loop save in start() existed only to persist team_readiness mutations from the resolver loop, so skip it when nothing changed. Restores the OCC contract of one version bump per start (test_occ_start_after_plan_save expects version 2); team plans still persist readiness diagnostics (covered by existing test_team_report_written_with_readiness_diagnostics, verified to fail if the save is removed). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…knowledge doctor hardening (bd-ftd) - api-reference: 422 plan_quality_error rows for all three forge routes (forge/plan and forge/regenerate were never documented after 3e67452) plus a shared Plan Quality Errors subsection with the detail shape - cli-reference: new Diagnostics section for baton doctor (previously undocumented) incl. plan_selection semantics and BATON_DB_PATH; knowledge doctor unreadable-pack, non-.md shadow rule, --knowledge-root de-dup - doc-audit: append 2026-07-06 sync entry Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…bt clearance Resolves sqlite_backend.py as the union of manager-mode (manager_mode column) and the fix wave's ON CONFLICT(task_id) DO UPDATE upsert that preserves release_id; extends migrate.py's legacy plans INSERT with manager_mode for parity. doc-audit.md keeps both appended entries. Storage/release/manager suites green post-merge (71+82+9+150). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
FallbackClassifier's talent-agent probe constructs HeadlessClaude during classification regardless of BATON_PLAN_REVIEW, colliding with these tests' HeadlessClaude mocks. Pin task_classifier=KeywordClassifier(), matching the pattern documented in tests/e2e/test_manager_mode_planning.py. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
… just injection (bd-pz4 part 1) The classified_phases Review/Audit guarantee fired only when RiskStage itself injected a safety agent; a pre-rostered reviewer/auditor produced no phase slot, force-landed in Implement, and was filtered — tripping the review_missing/audit_missing gates. Mirror the gate's own presence predicate (REVIEWER_AGENTS from router, same source as validation.py). compliance-audit golden regenerated: auditor now lands in a proper Audit phase instead of leaving a vestigial single-member team step. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
… part 2) The explicit-complexity path built phases via KeywordClassifier._select_phases (strips Review at medium) but kept the static DEFAULT_AGENTS roster (includes code-reviewer), producing reviewer-on-roster/no-Review-phase plans that the review_missing gate correctly hard-blocks. Filter reviewer-class agents (keeping auditor) from draft.resolved_agents when the override phase list has no Review phase — condition is phase-presence-driven, so heavy complexity keeps both. Gate, enrichment net, and fan-out guard untouched. Fixes both tests/e2e/test_manager_mode_planning.py failures; e2e 18/18. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
…actions Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WkNUh7JHsR4yJ6KTEhNmbm
… bd-b6f)
Two bugs found while standing up the pmo-ui e2e suite:
- SpecsPanel white-screened the ENTIRE app at mount: GET /api/v1/pmo/specs
is owned by the spec-queue router and returns a bare
list[SpecDraftResponse], but loadSpecs did setSpecs(res.specs) —
undefined — and the next render's specs.find threw with no error
boundary above, leaving #root empty at /pmo/. Coerce the response
shape defensively; regression test covers both shapes. The deeper
product question (SpecsPanel's Spec entity has no backing endpoint)
stays open in bd-091's notes; suite-wide goto('/') baseURL bug is
tracked separately in bd-m0b.
- e2e/utils/{screenshots,audit-reporter}.ts resolved their output dirs
via new URL(...).pathname, which on Windows yields "/C:/…" and
path.resolve mangles it into "C:\C:\…" — mkdirSync threw at import
time and killed the whole Playwright run. Use fileURLToPath.
Verified: vitest 12/12 (2 new); rebuilt dist; headless probe against
live `baton serve` now renders the full board (brand visible,
rootLen 165958, zero page errors) where it previously rendered a
blank page.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VMHTtWVqejqqnSQFGXsnLL
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Found while running the e2e sweep after the PR #109 landing:
SpecsPanelcrashed the entire PMO UI at mount — blank page at/pmo/. The spec-queue router ownsGET /api/v1/pmo/specsand returns a barelist[SpecDraftResponse], but the panel expected{specs: [...]}; the resultingundefined.findthrew during render with no error boundary. Fixed with defensive shape coercion + regression tests (both shapes).new URL(...).pathname→C:\C:\…on Windows, killing the whole suite at import. Fixed withfileURLToPath.Still open: bd-m0b — specs
goto('/')escapes the/pmo/baseURL path, so the Playwright suite (936/939 failing, not in CI) needs a navigation fix before it's meaningful.Verification
baton serve: board renders fully (brand visible, zero page errors) vs. blank page before.🤖 Generated with Claude Code
https://claude.ai/code/session_01VMHTtWVqejqqnSQFGXsnLL