beifeng724/personal-workbench-skills / skills/personal-workbench-builder
personal-workbench-builder
VerifiedBuild a durable personal or professional workbench as a runnable application, or produce a six-stage generic prompt pack for building it. Use when the user wants to turn a recurring workflow into a dashboard, management system, workspace, personal operating system, or long-lived website with multiple connected modules and persistent state. Trigger for project, CRM, creator, learning, family, finance, travel, teaching, legal, or team workbenches. Do not trigger for marketing sites, one-off scripts, ordinary coding tasks, or informational questions.
references/interview-and-levels.md
View on GitHub# Interview, Classification, And Confirmation
## Progressive interview
Ask only what is still unknown and group no more than three related questions at a time. Start with:
1. Output: runnable project or six-stage prompt pack?
2. Repeated workflow: what happens repeatedly, who does it, and what currently hurts?
3. Persistent information: what must remain available tomorrow, next month, or on another device?
Then cover, as relevant:
- Devices, mobile context, and mobile priority.
- Data sensitivity and acceptable storage location.
- Whether "use on all my devices" is needed. In plain mode, explain: "Do you want changes made on your phone to appear automatically on your computer?"
- Data importance and acceptable loss. In plain mode, ask whether recreating everything would be easy, painful, or unacceptable.
- Number of users and whether each person must see different data.
- Budget, including acceptable monthly recurring cost.
- Real-time data, integrations, scheduled work, and external APIs.
- Expected lifetime, future extensions, import/export, backup, and recovery.
- UI reference screenshot; otherwise audience, visual direction, expression,
motion, density, formality, theme mode, and mobile priority.
After gathering the visual evidence, state a one-sentence visual read and ask
the user to correct it only when the direction is materially uncertain. Do not
ask the user to understand framework or component-library names.
Do not ask the user to choose databases, hosting services, or frameworks unless they are technical and want that control. Translate the consequence first.
## Decisions required before specification approval
For every direct build, record all of these before `spec-approved`:
1. Primary devices and whether phone use is local, same-network, or anywhere.
2. Synchronization expectation: none, manual transfer, or automatic cross-device sync.
3. Data importance: replaceable, important, or hard to replace.
4. Acceptable data loss and acceptable recovery effort.
5. Persistence approach, backup owner, restore path, and deployment profile.
Do not infer automatic synchronization from "mobile support", a hosted URL, or installation on more than one device.
## Communication modes
- `plain`: The user explicitly says they are a beginner or consistently asks for non-technical explanations. Use everyday consequences and short steps. Define unavoidable terms on first use.
- `guided`: Default. Pair a plain-language consequence with the necessary technical label.
- `technical`: Use only after the user consistently demonstrates technical fluency. State architecture, trade-offs, security boundaries, and operating assumptions directly.
A single technical word is not enough evidence for `technical`. Ask which style the user prefers if the difference would affect consent or operation.
## Complexity levels
| Level | Deciding evidence | Typical architecture | Confirmation |
|---|---|---|---|
| L1 | One person, one device, no login, no cloud sync | Static/local app with browser or local-file storage | Fast path when requirements are clear |
| L2 | Durable URL, mobile use, import/export, or a public read-only API; data may remain in one browser | Hosted frontend, local-first persistence, or client-side public reads | Approve spec before development |
| L3 | Login, multi-device sync, cloud database/storage, scheduled/background jobs, credentialed APIs, write APIs, critical integrations, or legacy unspecified external APIs | Hosted frontend plus a trusted backend boundary | Approve spec, costs, privacy, recovery |
| L4 | Multiple users with roles/isolation, audit, notifications/payment, monitoring, backups, or ongoing operations | Operated business system | Repeat complete approval loop |
Choose the highest level implied by any confirmed requirement. Profession and module count do not decide the level.
Classify integrations by consequence:
- `public_readonly_api`: public, read-only, no secret, no privileged operation -> L2 unless another signal is higher.
- `credentialed_api`, `write_api`, `critical_integration`, or `scheduled_jobs` -> L3.
- Legacy `external_api` remains L3 because its risk is unspecified.
- Payment, multi-user roles/isolation, audit, monitoring, or ongoing operations remain L4 signals.
## Scope validation and ordering
Build a requirement ledger with statuses: `confirmed`, `proposed-by-ai`, `conflict`, `blocked`, `approved-for-later`, `rejected-by-user`.
- Preserve every `confirmed` item.
- Remove `proposed-by-ai` extras freely when they do not add clear value.
- Present dependencies as implementation order, not as feature deletion.
- For a conflict or blocked item, explain the consequence and offer alternatives.
- Move an item to later only when the user explicitly approves that decision.
- Include all later items in the specification and handoff context.
## Confirmation language
For L2-L3 ask: "This is the current complete specification. Do you approve it for implementation, and is there anything you want to add or change?"
For L4, summarize these six headings every cycle:
1. Features and workflows.
2. Data, retention, import/export, and backups.
3. Users, roles, permissions, and audit.
4. Deployment, monitoring, recovery, and ownership.
5. One-time and recurring costs.
6. Other ideas, future plans, and unresolved decisions.
Record the approval quotation and timestamp in `workbench-spec.md`. Any material change after approval reopens the relevant section and requires approval again.
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: 472k | Updated | Claude CodeCodex | Unverified | 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 |