chore(release): version packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and publish to npm yourself or setup this action to publish automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

rn-dev-agent-cdp@0.45.1

Patch Changes

• bd5d585: Recovery paths now detect "app not installed" and resolve their relaunch target truthfully (GH bug: cdp_status APP_DETACHED recovery doesn't detect "app not installed" #262, absorbs iOS verification UX: stale sessions, agent-device/rn-fast-runner conflict, destructive clearState, and recovery loops #194 BUG 2).

  • cdp_status APP_DETACHED auto-relaunch: when simctl launch fails AND get_app_container's stderr carries the NSPOSIXErrorDomain code=2 marker (allowlist-only, stderr-only — argv-spoof-proof), the tool returns a distinct APP_NOT_INSTALLED code with install advice — including a shell-quoted simctl install line for the newest matching .app snapshot from the last clearState (GH bug: maestro_run can't pass --app-file, so clearState:true flows on iOS force a raw-CLI escape #201 dir, mtime-sorted budgeted scan). Ambiguous probe verdicts fail open to the existing APP_DETACHED behavior. Concurrent recoveries are serialized, and a confirmed missing bundle is cached (with a cheap re-probe) so the diagnosis is never masked by budget-exhausted.
  • cdp_restart hardReset=true: the relaunch target resolves through explicit arg > connectedTarget > cache > active-session appId > strict per-platform app.json (no iOS←Android fallback), simctl targets the active session's UDID when one exists, failed launches are classified the same way in hardResetSteps, and a successful hard reset resets the detached-recovery budget.

• 81c386a: device_screenshot no longer blames "device transitioning state" when the target directory doesn't exist (GH bug: device_screenshot blames "device transitioning state" when the target directory doesn't exist #265).

  • captureAndResizeScreenshot now mkdir -p's the parent of the derived output path before any dispatch tier runs (simctl raw, rn-fast-runner, agent-device daemon/CLI, adb stream) — new directories are the expected case, since the tool's own advisories steer agents toward fresh docs/proof/<slug>/ paths. The fix covers device_screenshot, device_batch auto-captures, and proof_step, all of which funnel through the same helper.
  • When the directory itself cannot be created (e.g. a file blocks an intermediate path segment), the tool short-circuits before probing any device and returns an honest SCREENSHOT_FAILED with reason: 'target-dir-unavailable' naming the offending path — never the device-state guess.
  • A leading ~/ in the screenshot path is now expanded to the real home directory (Node never expands ~, so mkdir would otherwise create a literal ./~/ under the bridge cwd and report success into the wrong location). Unexpandable forms (~user/..., bare ~) are refused with an actionable error.

rn-dev-agent-plugin@0.54.1

Patch Changes

• bd5d585: Recovery paths now detect "app not installed" and resolve their relaunch target truthfully (GH bug: cdp_status APP_DETACHED recovery doesn't detect "app not installed" #262, absorbs iOS verification UX: stale sessions, agent-device/rn-fast-runner conflict, destructive clearState, and recovery loops #194 BUG 2).

  • cdp_status APP_DETACHED auto-relaunch: when simctl launch fails AND get_app_container's stderr carries the NSPOSIXErrorDomain code=2 marker (allowlist-only, stderr-only — argv-spoof-proof), the tool returns a distinct APP_NOT_INSTALLED code with install advice — including a shell-quoted simctl install line for the newest matching .app snapshot from the last clearState (GH bug: maestro_run can't pass --app-file, so clearState:true flows on iOS force a raw-CLI escape #201 dir, mtime-sorted budgeted scan). Ambiguous probe verdicts fail open to the existing APP_DETACHED behavior. Concurrent recoveries are serialized, and a confirmed missing bundle is cached (with a cheap re-probe) so the diagnosis is never masked by budget-exhausted.
  • cdp_restart hardReset=true: the relaunch target resolves through explicit arg > connectedTarget > cache > active-session appId > strict per-platform app.json (no iOS←Android fallback), simctl targets the active session's UDID when one exists, failed launches are classified the same way in hardResetSteps, and a successful hard reset resets the detached-recovery budget.

• 81c386a: device_screenshot no longer blames "device transitioning state" when the target directory doesn't exist (GH bug: device_screenshot blames "device transitioning state" when the target directory doesn't exist #265).

  • captureAndResizeScreenshot now mkdir -p's the parent of the derived output path before any dispatch tier runs (simctl raw, rn-fast-runner, agent-device daemon/CLI, adb stream) — new directories are the expected case, since the tool's own advisories steer agents toward fresh docs/proof/<slug>/ paths. The fix covers device_screenshot, device_batch auto-captures, and proof_step, all of which funnel through the same helper.
  • When the directory itself cannot be created (e.g. a file blocks an intermediate path segment), the tool short-circuits before probing any device and returns an honest SCREENSHOT_FAILED with reason: 'target-dir-unavailable' naming the offending path — never the device-state guess.
  • A leading ~/ in the screenshot path is now expanded to the real home directory (Node never expands ~, so mkdir would otherwise create a literal ./~/ under the bridge cwd and report success into the wrong location). Unexpandable forms (~user/..., bare ~) are refused with an actionable error.