saas-kit
Skill14 starsChanged 5 months agoMITView on GitHub
Full Spec Kit-inspired workflow for AI-assisted SaaS planning, implementation, review, and evolution. Use when the user wants to plan, spec, build, audit, or evolve a SaaS product, MVP, dashboard, CRUD app, workflow system, admin tool, PocketBase/Supabase-backed app, or Spec Kit artifact set. The skill enforces one A/B/C/D clarification question at a time, produces a detailed SaaS spec focused on product workflows before stack decisions, maps the result to constitution/spec/plan/tasks/checklists/analyze/converge artifacts, supports Spec Kit when installed or a manual equivalent when absent, and blocks coding until the user approves the spec.
In this repository12 other files it gives its agents.
- brandkitSkill
- dublyobase-backend-builderSkill
- dublyo-deploySkill
- image-to-codeSkill
- lean-codeSkill
- mdn-web-best-practicesSkill
- moatSkill
- playwright-website-testerSkill
- dublyo-pocketbaseSkill
- saas-product-growthSkill
- dublyo-tag-manager-serverSkill
- design-taste-frontend-v1Skill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 12 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Do not write or edit implementation code until the user has approved a spec.
- Distinctive: line 18
Create or update a plan, checklist, and tasks only after the spec is clear.
- Distinctive: line 19
Ask for explicit approval before implementation.
- Distinctive: line 20
Only after approval, inspect or modify implementation files.
- Distinctive: line 23
If the user asks for a review or audit, stay read-only unless they explicitly ask for fixes.
- Distinctive: line 27
Ask at most 10 questions.
- Distinctive: line 28
Ask exactly one question per message.
- Distinctive: line 32
Stop early when the product direction is clear enough to write a useful spec.
- Distinctive: line 68
After the question loop, write a detailed spec with these sections:
- Distinctive: line 89
If a repo has Spec Kit installed, use its artifact model and commands where appropriate:
- Distinctive: line 99
If Spec Kit is absent, follow the same workflow manually using Markdown specs and task lists.
- Distinctive: line 101
Treat these artifacts as the source of intent:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
SaaS Kit
Operating Rule
Do not write or edit implementation code until the user has approved a spec.
Use this order:
- Establish or note the SaaS constitution and project rules.
- Clarify the SaaS product with one question at a time.
- Produce a full human-readable SaaS spec.
- Map the spec to Spec Kit/manual artifacts.
- Create or update a plan, checklist, and tasks only after the spec is clear.
- Ask for explicit approval before implementation.
- Only after approval, inspect or modify implementation files.
- Analyze before implementation and converge after implementation.
If the user asks for a review or audit, stay read-only unless they explicitly ask for fixes.
Question Protocol
- Ask at most 10 questions.
- Ask exactly one question per message.
- Each question must have exactly four choices: A, B, C, D.
- Each question must build on the previous answer.
- Recommend one option when helpful, but still present all four.
- Stop early when the product direction is clear enough to write a useful spec.
Prioritize questions in this order:
- Primary SaaS job and target user
- Core workflow and success moment
- Dashboard views and daily user actions
- CRUD objects and ownership rules
- Roles and permissions
- Workflow states and transitions
- Data/provider constraints
- UI density and operational needs
- Testing and verification expectations
- Implementation scope and first release boundary
Planning Bias
Focus on the app's real work, not marketing copy or premature architecture.
Default assumptions:
- Product: core SaaS workflows, dashboards, CRUD, and business functions first.
- Frontend: functional dashboard-first UI, Tailwind CSS by default.
- Backend/data: provider-neutral first; map to PocketBase or Supabase after workflows are clear.
- New app defaults: Next.js, Nuxt.js, or Astro.js for frontend; PocketBase or Supabase for backend/data.
- Existing app: inspect repo conventions after spec approval and follow existing patterns.
Avoid:
- Landing-page-first planning unless requested.
- Billing, subscriptions, enterprise compliance, or SSO by default unless central to the product.
- Decorative UI that hides workflow clarity.
- New abstractions or dependencies before a real need is visible.
Spec Output
After the question loop, write a detailed spec with these sections:
- Goal
- Scope
- User flow
- Requirements
- Frontend
- Backend
- Database
- APIs
- UI/UX
- Edge cases
- Security
- Testing
- Spec Kit/manual artifact mapping
- Implementation steps
Then ask: "Approve this spec for implementation?" Do not code until the user approves.
Spec Kit Use
If a repo has Spec Kit installed, use its artifact model and commands where appropriate:
- Constitution/project rules
spec.mdplan.mdtasks.md- requirement checklists
analyzebefore implementationconvergeafter implementation
If Spec Kit is absent, follow the same workflow manually using Markdown specs and task lists.
Treat these artifacts as the source of intent:
constitution.md: non-negotiable project principles and gatesspec.md: user intent, scenarios, requirements, entities, edge cases, success criteriaplan.md: technical context, provider mapping, contracts, validation guidetasks.md: user-story-sized work, exact files, dependencies, checkpointschecklists/*.md: "unit tests for English" that validate requirement qualityanalyze: read-only consistency report before implementationconverge: append-only remaining-work discovery after implementation
Reference Routing
Load only the reference needed for the current task:
references/workflow-spec.md: question loop, spec template, and Spec Kit/manual mapping.references/spec-kit-lifecycle.md: full constitution -> specify -> clarify -> plan -> checklist -> tasks -> analyze -> implement -> converge flow.references/constitution-governance.md: SaaS constitution principles, approval gates, and governance rules.references/artifact-lifecycle.md: spec/plan/tasks relationships, persistence models, drift handling, and monorepo notes.references/dashboard-crud.md: dashboard-first UI, CRUD, workflow states, and SaaS screen planning.references/pocketbase-supabase.md: provider-neutral mapping to PocketBase or Supabase.references/testing-verification.md: quality gates, tests, smoke checks, and approval/implementation verification.references/implementation-convergence.md: task generation, scoped implementation, subagent use, and convergence rules.references/customization-integration.md: presets, extensions, bundles, Git hooks, and agent integration guidance.
For web implementation, follow MDN Web Docs best practices and the repo's existing patterns. When library/API syntax matters, fetch current official docs before coding.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

