On the dependency-guard check: stacked on the not-yet-merged Wave 1 (#101098), inheriting its single new-plugin pnpm-lock.yaml importer entry. No new dependency. Verified by the submitting orchestrator's own gate + security review: node scripts/test-projects.mjs extensions/dashboard 89 passed, vitest src/lib/dashboard 103 passed, ui:build + full type-aware oxlint clean. Security: stream is an allowlist over existing gateway broadcast events (no new socket); computed is a fixed op switch (no eval); the command: binding is deliberately deferred to its own reviewed PR.

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

Dependency graph guard cleared

This PR no longer has blocked dependency graph changes. A future dependency graph change requires a fresh /allow-dependencies-change comment after the guard blocks that new head SHA.

• Current SHA: 68d4869fc655e41183bc451748fcf4e7e75dcadd

Codex review: needs real behavior proof before merge. Reviewed July 13, 2026, 8:37 AM ET / 12:37 UTC.

Summary
Adds allowlisted live-event and fixed-operation computed bindings to Workspaces, explicit widget output selection, persisted-document migration, authoring support, UI resolution, tests, and documentation.

PR surface: Source +662, Tests +515, Docs +24. Total +1201 across 19 files.

Reproducibility: not applicable. This PR adds new Workspaces capabilities rather than fixing a reproducible current-main bug. Source and tests establish the proposed behavior, but not an existing failure.

Review metrics: 2 noteworthy metrics.

• Binding contract surface: 2 binding kinds added, 1 output selector added. These become authoring, persistence, gateway, and UI contracts rather than an isolated rendering change.
• Persisted migration paths: 1 on-load document rewrite added. Existing multi-binding workspace documents can be modified during ordinary runtime loading, requiring explicit upgrade proof and ownership.

Stored data model
Persistent data-model change detected: database schema: ui/src/pages/plugin/workspace-view.test.ts, serialized state: extensions/workspaces/src/store.ts, unknown-data-model-change: extensions/workspaces/src/schema.test.ts, unknown-data-model-change: extensions/workspaces/src/schema.ts. Confirm migration or upgrade compatibility proof before merge.

Root-cause cluster
Relationship: fixed_by_candidate
Canonical: #101150
Summary: This PR is a candidate partial implementation of the canonical binding-kinds issue, covering stream and computed bindings while deliberately leaving command bindings unresolved.

Members:

• canonical: Dashboard: new data-binding kinds — stream, command, computed #101150 - It directly requests stream, command, and computed binding kinds and is named by the PR's closing reference.
• partial_overlap: Modular Dashboard — Wave 2+ roadmap (tracking) #101136 - The roadmap tracks this PR as the binding-kinds portion of the broader Workspaces initiative.
• adjacent_distinct: Dashboard: live log-tail + command-output widgets #101143 - Live log and command-output widgets could consume future stream bindings, but they require distinct widget and gateway capabilities.
• adjacent_distinct: feat(workspaces): add agent-composable Workspaces #104139 - That merged PR established the current Workspaces foundation on which this separate binding extension is built.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Merge readiness
Overall: 🦪 silver shellfish
Proof: 🦪 silver shellfish
Patch quality: 🐚 platinum hermit
Result: blocked until real behavior proof is added.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

• [P1] Add a redacted recording or screenshots with diagnostic/live output showing stream updates and computed rendering in a real Control UI session.
• Show an existing SQLite workspace with multiple bindings upgrading, reloading, undoing, and rendering without data or output-selection drift.
• Obtain Workspaces-owner approval for the event allowlist, migration ownership, and narrower canonical-issue scope.

Proof guidance:

• [P1] Needs real behavior proof before merge: The PR provides tests and CI but no inspectable after-fix Workspaces recording/live output or existing-SQLite upgrade proof; add redacted UI evidence and an upgrade transcript, update the PR body to trigger review, or ask a maintainer for @clawsweeper re-review.

Risk before merge

• [P1] The on-load migration rewrites existing workspace documents and changes multi-binding interpretation; tests do not demonstrate a real existing-SQLite upgrade, undo, reload, and subsequent render cycle.
• [P1] Allowing declarative widgets to consume gateway broadcasts expands the data-flow boundary into trusted UI and potentially approved custom-widget rendering, so the exact event allowlist needs explicit owner acceptance.
• [P1] Closing Dashboard: new data-binding kinds — stream, command, computed #101150 with this PR would hide the still-unimplemented command-binding portion of that request.
• [P1] The branch is behind current main, so its exact-head review and proof should be refreshed after rebasing even though the supplied state reports it as mergeable.

Maintainer options:

1. Narrow, approve, and prove (recommended)
   Retain stream/computed support, correct the canonical-issue closure scope, settle the migration owner, and add real Control UI and existing-database upgrade evidence before merge.
2. Accept the runtime migration
   A Workspaces owner may explicitly accept on-load document rewriting and the event allowlist after reviewing the upgrade and data-exposure consequences.
3. Defer the binding expansion
   Pause or close the PR if the permanent binding API and event-exposure contract are not currently sponsored.

Next step before merge

• [P1] The contributor should add real UI and upgrade proof, then the Workspaces owner must decide the event-exposure, migration, and partial canonical-issue contracts; no narrow autonomous code repair is currently established.

Maintainer decision needed

• Question: Should Workspaces permanently expose allowlisted gateway events and client-side computed operations as a persisted binding contract, while leaving command bindings for separate review?
• Rationale: The choice expands a plugin API and browser data-flow boundary, changes persisted workspace interpretation, and only partially fulfills the canonical issue; tests cannot determine whether that permanent product and upgrade contract is desired.
• Likely owner: steipete — The current Workspaces architecture and its persistence, gateway, and UI boundaries originate from the merged integration work associated with this owner.
• Options:
  • Sponsor the narrow contract (recommended): Approve stream and computed bindings only, keep command bindings separately tracked, require an owner-approved migration shape, and obtain real UI and upgrade proof before merge.
  • Complete the canonical scope: Hold this PR until command bindings have an explicitly approved security model and the complete canonical request can be reviewed together.
  • Pause the feature: Close or defer this PR and keep the canonical issue open until Workspaces binding extensibility becomes a confirmed roadmap priority.

Security
Cleared: No concrete exploit or supply-chain regression is evident: streams use a closed read-event allowlist, computed operations avoid arbitrary evaluation, and the diff adds no dependency or executable download path.