TASK-001E: Streaming, Diagnostics, and Defects

• Status: Ready
• Parent: TASK-001
• Spec: SLICE-001
• Depends on: TASK-001C, TASK-001D

Goal

Add bounded telemetry and preview frame flow, simple defect generation, and enough diagnostics visibility to understand the runtime without overloading the UI or memory.

Scope

• implement bounded telemetry and preview frame pipelines
• implement coalescing or recency-preserving lag policy
• surface diagnostics or counters for dropped/coalesced work
• implement simple defect generation tied to the current run
• feed diagnostics and stream outputs into canonical app state

Non-Scope

• advanced image processing
• advanced metrics dashboards
• persistent diagnostics history

AI Execution Profile

• Model class: High-capability or Balanced
• Reasoning effort: High
• Prefer the stronger model if your tool struggles with concurrency or cancellation

Files Or Areas Expected To Change

• Infrastructure producers for telemetry and frames
• Application stream consumption and diagnostics logic
• defect generation path and counters

Acceptance Targets

• telemetry and preview pipelines are bounded
• lag policy is explicit and testable
• defect count updates are associated with the current run
• diagnostics make dropped/coalesced behavior observable

Copy/Paste Prompt

Implement only TASK-001E: Streaming, Diagnostics, and Defects in this repository.

Read first:
- docs/requirements.md
  Focus on Sections 7.6, 7.7, 7.8, 7.11, 9.6, 9.7, 9.8, 9.9, and 13.3
- docs/specs/SLICE-001-first-strong-vertical-slice.md
  Focus on Streaming Policy, Run Execution, and Verification Notes
- docs/tasks/slice-001/TASK-001E-streaming-diagnostics-and-defects.md

Goal:
- Implement bounded telemetry and preview frame flow, simple defect generation, and diagnostics visibility for the first slice.

Scope:
- Implement bounded pipelines for telemetry and preview frames.
- Use explicit lag behavior:
  telemetry may coalesce to latest value
  preview frames should prefer recent frames over stale backlog
- Make dropped/coalesced behavior observable through diagnostics, counters, or logs.
- Generate simple defects and associate them with the current run.
- Feed stream-derived state into the canonical app state path rather than updating UI objects directly.

Do not:
- Add advanced image processing
- Add full analytics dashboards
- Create unbounded in-memory queues
- Bypass the canonical state path

Verification:
- Build if possible
- Add or update tests for bounded/coalesced behavior if practical
- Summarize the lag policy and how it is enforced

Important:
- Implement only this task
- Do not preemptively implement later tasks