| name | flutterhelm-stable-flow |
| description | Guide stable FlutterHelm setup and execution through the default npx-backed plugin workflow. Use when Codex needs to start or operate stable FlutterHelm flows such as workspace selection, run and inspect, tests, profiling, compatibility checks, or native handoff without enabling beta runtime_interaction or native_build paths. |
FlutterHelm Stable Flow
Use this skill to run the default, stable FlutterHelm loop through the plugin's bundled npx wrapper.
Workflow
- Start from the default plugin MCP config and treat
stdio as the primary path.
- Confirm readiness with
workspace_show, compatibility_check, and adapter_list.
- Set or discover the workspace root before opening or running a session.
- Use the stable loop:
run_app
get_logs
get_runtime_errors
get_widget_tree
get_app_state_summary
capture_screenshot
- Move to
run_unit_tests, run_widget_tests, run_integration_tests, or profiling tools only after the session/root state is healthy.
- Use
ios_debug_context, android_debug_context, or native_handoff_summary when Flutter-side diagnosis is not enough.
Default expectations
- The plugin default path uses
npx -y github:CAPHTECH/flutterhelm serve.
- Prefer the built-in delegate and let FlutterHelm handle official Flutter MCP fallback automatically.
- Keep
runtime_interaction and native_build disabled unless the user explicitly asks for beta behavior.
- Prefer
session://<id>/health, config://compatibility/current, and config://adapters/current over guessing why a workflow is unavailable.
References
- Read
references/stable-commands.md for the default MCP config, setup commands, and the stable tool sequence.
- Read
../../../../docs/11-user-guide.md when you need the repo's full user-facing walkthrough.