[BIackFIame]

Summary

Part 4/7 of the CanvasTTY agent-workflow series. Durable timelines, session reports and searchable output history explain long-running work. Missing usage remains unknown; loop-detection hooks and attention events expose stalled work. Output history includes streaming secret redaction and masked worker snapshots.

Review scope and dependencies

Depends on: #133.

This is a cumulative upstream PR: GitHub's Files changed tab includes earlier unmerged parts because all PRs target upstream main; cross-fork stacked bases are unsupported. Review this part in the isolated predecessor-to-head diff. The intended merge order follows the numbered series; do not merge a later part independently of its predecessors.

The Assistant plugin supplies loop judgments. Phone transport is in the final experimental part.

Validation and current blockers

Tested feature head: 21d5c99f7957ec98ea5aa893127ec6360b6084fe. Local helpers, typecheck, build and secret audit passed; full suite: 1,968 passed, 8 skipped, zero failures. Historical fork CI passed on macOS, Linux and Windows using the exact candidate tree under a temporary workflow overlay. That overlay is not included here.

Draft: merge preview against updated upstream 6faadcdc found conflicts in src/preload/index.ts, src/renderer/src/features/workspace/WorkspaceCanvas.tsx, src/shared/contracts.ts. These historical results do not validate a conflict-resolved integration with current main. Preserve the newly accepted material/terminal-link behavior, resolve the overlaps, and rerun checks before marking ready.

Historical hidden-app evidence recorded timeline events and notification payloads; actual OS/phone delivery was not exercised. A deterministic loop fixture triggered after six denied actions, without model calls. Real-model loop detection and large-history performance remain unverified.

Series order

1. macOS clipboard
2. Orchestration, reviews and budgets — Draft
3. Worktrees and checkpoints — Draft
4. Timeline, usage and attention — Draft
5. Network policies and temporary secrets — Draft
6. Workspace context and navigation — Draft
7. Experimental routing, handoff and remote controls — Draft

Parts 2–7 are cumulative Draft PRs against upstream main, with an isolated layer diff in each description. Their tested feature heads currently conflict with newer upstream changes; integration and fresh checks are required before readiness. Part 7 also retains its independent experimental acceptance gate.

[BIackFIame]

@codex review

Please review this PR for correctness, regressions and security issues. Review only; post findings on GitHub without changing code.

This is part 4/7 of a cumulative series. Focus on changes introduced after predecessor 2b988218680596776177862ef2fb53daa5549c6b through current head 21d5c99f7957ec98ea5aa893127ec6360b6084fe, including their interactions with the inherited code. The isolated layer diff is linked in the PR description. Known merge conflicts with newer main are already documented; look for substantive code defects in addition to those integration blockers.

[howdeploy]

@BIackFIame, we are rejecting this PR in its current form and closing it following the maintainer's local DEV acceptance. The reasons are the same product and design concerns already explained for #132 and #137: the added interface does not follow CanvasTTY's design contract and adds too much harness administration to the default canvas experience.

Related feedback:

• #132
• #137

Design and interaction

The maintainer inspected the session-activity interface in the combined Linux DEV build and rejected its composition and complexity. This is a product acceptance decision, not a claim that every operating system has the same rendering failure.

There are concrete implementation problems behind that feedback. The inherited BacklogSessionInspector is extended into a multi-section administration surface, with timeline filters, usage breakdowns, reports and notification preferences alongside the task, budget and checkpoint controls from preceding parts. Its custom stylesheet fixes the inspector background to #252733, uses white text and independently styled controls, and adds multiple fixed-pixel font sizes and spacing values. Those choices do not consistently participate in our palette roles and root UI scale. The close action is a literal multiplication character rather than the existing UiIcon close asset. Timeline event types and sources are displayed directly from backend values instead of through localized presentation labels.

Please use the existing UI contract, components, theme tokens, typography, icons and interaction patterns for any future contribution. Recoloring this inspector will not solve the larger UX problem: normal canvas use should not require understanding another set of event filters, usage categories, price settings and notification channels. Optional diagnostics need a clear purpose, a discoverable entry and deliberate opt-in.

Scope and maintenance

The #133-to-#134 layer includes durable timelines, searchable terminal-output storage, Markdown session reports, usage attribution and pricing, attention preferences and loop-detection integration. It also changes provider runtime hooks, the native helper, terminal/session lifecycle, redaction, worker processes, IPC/preload and companion presentation. This is a substantial set of independently reviewable concerns, not just an additional report screen.

The architectural overlap matters even when Git can combine the files. Other contributions use the same terminal input/output, session lifecycle, plugin-event and usage paths. Each additional observer, persistent history and completion report must preserve ordering, cancellation, secret handling and restart behavior. A successful build or historical CI run does not establish product acceptance or compatibility across those combined paths.

Requested direction

Please move the activity dashboard, report presentation, usage/pricing administration, notification configuration and loop-analysis workflow into an optional CanvasTTY plugin. Start with the existing supervised services, session events, permission-controlled screen access and plugin UI contributions described in ARCHITECTURE.md.

If the plugin needs an operation that those APIs cannot express, propose a narrowly scoped core extension separately, with the concrete use case, data contract, permissions, retention, cancellation and disabled-plugin behavior. Authoritative validation, secret redaction and privileged operations must remain enforced by the trusted host. Independently useful runtime or redaction fixes may be proposed as focused core PRs; they should not require accepting this entire UI and harness feature bundle.

This series is cumulative. #134 includes earlier unmerged work, including the rejected #132, and later parts include #134. Please split or rebase remaining proposals so rejected functionality cannot return through another cumulative PR. Closing this PR rejects the current scope; it does not approve that scope through its successors.