Orchestration Planning Ux

[arul28]

Summary

Describe the change.

What Changed

Key files and behaviors.

Validation

How you tested.

Risks

Anything to watch.

  Open in ADE  ·  ade/orchestration-planning-ux-88fff44d branch  ·  PR #605

Summary by CodeRabbit

Release Notes

• New Features
  • Rolled out a deterministic lead planning workflow with structural plan readiness checks, content-hash–gated approvals, and ordered unlocking of model selection.
  • Enhanced model-selection prompts/cards with “Files it touches” and “Runs after” briefing chips and updated UI wording.
  • Added a planning timeline and a validation findings panel with severity ordering and regression-test indicators.
  • Improved orchestration chat navigation (worker/task open routing) and added a plan-approval footer that flags changes since last review.
• Documentation
  • Updated orchestrator planning/validation protocol documentation.
• Tests
  • Expanded orchestration gating, UI, and approval/rejection coverage (including metadata-driven briefing rendering).

Greptile Summary

• Adds a deterministic orchestration planning and approval workflow with gated intake, deliberation, validation, model selection, and plan approval states.
• Introduces structural plan.md readiness checks, approval hashing, re-approval handling, and model-selection briefing metadata.
• Updates desktop, TUI, and iOS orchestration surfaces with planning timeline, validation findings, approval, and model-selection UI changes.
• Expands orchestration gate, approval, model-selection, recovery, and UI test coverage.

Confidence Score: 5/5

No blocking code issues were identified in the reviewed changes.

The changes include focused coverage across orchestration gates, approval handling, model-selection state, recovery behavior, and UI surfaces, with no reported defects requiring code changes.

T-Rex Logs

What T-Rex did

• T-Rex ran the requested verification, but its local artifact references were not uploaded.

_{Ran code and verified through T-Rex}

Comments Outside Diff (3)

1. General comment

   Desktop model selection file row is labeled Touches instead of Files it touches

   • Bug
     • The new desktop briefing row for filesHint renders the label Touches. The validation contract requested the desktop card visibly show Files it touches (or Files for TUI). This may make the UI inconsistent with the intended prompt/card copy even though the metadata is otherwise consumed and truncated.
   • Cause
     • In apps/desktop/src/renderer/components/chat/ChatModelSelectionPendingCard.tsx, the file row label text is hard-coded as Touches.
   • Fix
     • Change the desktop file row label to Files it touches if that is the intended UX contract, leaving TUI as Files.

   _{Ran code and verified through T-Rex}

2. apps/desktop/src/main/services/orchestration/patchPolicy.ts, line 332-334 (link)

   Don’t pass blockers

   recordValidationRun can now attach structured findings, but this gate only checks that the latest run has status === "passed". A validator can record a passed run with a blocker or high finding, the Findings panel will still show the blocking issue, and releaseTask(status: "done") will allow the task to complete. Make the gate reject latest runs with unresolved blocker/high findings, or reject passed validation runs that contain them.

   Artifacts

   
   

   Repro: focused harness for passed validation run with blocker and high findings

   • Contains supporting evidence from the run (text/typescript; charset=utf-8).

   Stack trace captured during the T-Rex run

   • Keeps the raw stack trace available without making the summary code-heavy.

   

   _{Ran code and verified through T-Rex}

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: apps/desktop/src/main/services/orchestration/patchPolicy.ts
   Line: 332-334
   
   Comment:
   **Don’t pass blockers**
   
   `recordValidationRun` can now attach structured findings, but this gate only checks that the latest run has `status === "passed"`. A validator can record a passed run with a `blocker` or `high` finding, the Findings panel will still show the blocking issue, and `releaseTask(status: "done")` will allow the task to complete. Make the gate reject latest runs with unresolved blocker/high findings, or reject `passed` validation runs that contain them.
   
   How can I resolve this? If you propose a fix, please make it concise.

   

3. General comment

   Model-selection ApprovalPrompt drops orchestration briefing metadata

   • Bug
     • The changed prompt contract says model-selection briefings should visibly include metadata such as touched files and runs-after dependencies. Rendering the head ApprovalPrompt with providerMetadata containing workDescription, filesHint, and dependsOn produced the same visible output as base: only the model-selection title, the model question, and key hints. The expected Description: ..., Files: ..., and Runs after: ... lines were absent.
   • Cause
     • The question-rendering path in apps/ade-cli/src/tuiClient/components/ApprovalPrompt.tsx renders question header/body/options starting around the isQuestion && questions.length block, but the component does not read or render approval.request.providerMetadata fields such as workDescription, filesHint, or dependsOn before displaying the model-selection question.
   • Fix
     • Add model-selection-specific metadata extraction/rendering before the questions list, normalizing and trimming providerMetadata.workDescription/filesHint/dependsOn into visible lines, and ensure the focused ApprovalPrompt test asserts those lines against the rendered frame.

   _{Ran code and verified through T-Rex}

_{Reviews (6): Last reviewed commit: "ship: iteration 15 address validation ga..." | Re-trigger Greptile}