try-works/role-model / .agents/skills/recursive-mode
recursive-mode
UnverifiedRepository workflow orchestration skill for staged implementation, locked artifacts, late-phase receipts, and durable memory maintenance. Use when executing recursive-mode runs, resuming a run, locking a phase, or verifying locks.
references/benchmarks/local-first-planner/00-requirements.md
View on GitHub## Requirements
### `R1` Local planner work-item management
Description: Build a local-first planner app where a user can create, edit, and delete work items without any backend service.
Acceptance criteria:
- A user can create a work item with title, short description, status, priority, optional due date, and tags.
- A user can edit an existing work item and save the updated values.
- A user can delete a work item with a clear UI affordance.
- The UI keeps the item list and any summary widgets in sync after each change.
### `R2` Search, filters, and grouped planning view
Description: The planner should make it easy to review work items by search and filter state in a local browser session.
Acceptance criteria:
- A user can search work items by text.
- A user can filter work items by status and priority.
- The main UI presents work items in a grouped planner view that makes status visible at a glance.
- The UI includes summary counts that reflect the currently displayed data set.
### `R3` Local persistence and resettable sample data
Description: The app must work entirely in the browser and preserve data across reloads.
Acceptance criteria:
- The planner persists its data in browser-local storage only.
- Reloading the page restores the most recently saved planner state.
- The app provides a way to load or restore sample data for evaluation.
- The app provides a way to reset back to the packaged sample state without requiring a backend.
### `R4` Import and export of planner data
Description: The benchmark app should support basic portability of its local planner data.
Acceptance criteria:
- A user can export the current planner data set as JSON.
- A user can import a previously exported JSON data set into the app.
- Invalid imported JSON is rejected with a visible error state instead of silently failing.
- Importing data updates the grouped view and summary counts immediately.
### `R5` Local-browser UX states
Description: The app should feel finished enough for browser-based validation rather than only passing a build.
Acceptance criteria:
- The app includes a clear empty state when no work items are present.
- Overdue items are visually distinguishable from non-overdue items.
- The main layout remains usable at typical laptop-browser widths.
- The app exposes enough visible UI state that a browser screenshot can show the planner, filters, and summary content.
### `R6` Quality gates
Description: The benchmark result should be judged on working software rather than source changes alone.
Acceptance criteria:
- `npm run build` succeeds.
- `npm run test` succeeds.
- `npm run preview -- --host 127.0.0.1 --port <port>` can serve the built app locally.
- The implementation includes at least a small automated test surface for important planner behavior.
## Out of Scope
- `OOS1`: Authentication, accounts, or multi-user collaboration.
- `OOS2`: Any server, API, database, Docker, or cloud dependency.
- `OOS3`: Drag-and-drop kanban interactions.
- `OOS4`: Real-time synchronization across browser tabs or devices.
## Constraints
- Use React + TypeScript + Vite.
- Keep all state local to the browser.
- The project must run from a disposable temp folder.
- Do not require any external service beyond standard Node/npm package installation.
- Keep the app suitable for later browser-agent screenshot validation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |