fix(windows): make recovery decisions observable and CLI guidance visible

[zackees]

Refs #1152 / meta #1149. Two gaps found during live hardware closeout on the wedged board (serial 5303284720C4641C): the daemon composed the typed recovery request silently (no audit trail), and the CLI guidance was emitted at tracing WARN, which the default filter suppresses — the user saw no diagnosis and no --admin hint. Daemon now logs composed/none-composed decisions; CLI recovery messaging moved to output::diagnostic. Live-verified: default-policy deploy prints the exact devnode (problem 28) + guidance without elevating.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes

  • Improved USB recovery messaging during deployment by presenting diagnostic guidance at appropriate severity levels.
  • Added clearer recovery status details when a Windows device requires recovery, recovery is declined or cancelled, recovery fails, or no healthy connection is detected.

• Diagnostics

  • Improved deployment logging for RP2040 USB recovery decisions, including when recovery requests are created or unavailable.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 06b4d0a1-3b73-4d54-bc71-5c4601da92ea

📥 Commits

Reviewing files that changed from the base of the PR and between d0255fe and 69b9315.

📒 Files selected for processing (2)

• crates/fbuild-cli/src/cli/deploy.rs
• crates/fbuild-daemon/src/handlers/operations/deploy.rs

---

📝 Walkthrough

Walkthrough

The deploy recovery flow changes several CLI messages from warnings to diagnostics and adds structured logging around RP2040 recovery-request composition. Recovery and retry behavior remain unchanged.

Changes

USB recovery observability

Layer / File(s)	Summary
CLI recovery diagnostic output crates/fbuild-cli/src/cli/deploy.rs	Known-unhealthy device, incomplete recovery, helper failure, and missing CDC endpoint messages now use diagnostic output while preserving existing branch behavior.
RP2040 recovery request logging crates/fbuild-daemon/src/handlers/operations/deploy.rs	The composed RP2040 recovery request is stored locally, logged with its fields when present, or accompanied by fresh-scan counts when no eligible target is found. The request is then returned unchanged.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related issues

• FastLED/fbuild issue 1152: The daemon recovery-request composition and recovery outcome handling align with the issue’s RP2040 recovery orchestration.

Possibly related PRs

• FastLED/fbuild#1153: Overlaps directly in the RP2040 deploy recovery and port reacquisition path.
• FastLED/fbuild#1157: Updates the device enumeration used by the fresh scan that feeds recovery-request eligibility.
• FastLED/fbuild#1068: Uses diagnostic output for recovery-related CLI messaging in the same deploy output path.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/1152-recovery-observability

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}