Project Templates

README.md Template

# [Project Name]

## What We're Building
[One paragraph description of the project and its purpose]

## Current Phase
[MVP / Sprint X / Beta / Production]

## Tech Stack
- Frontend: [framework]
- Backend: [approach]
- Database: [database + ORM]
- Auth: [auth strategy]
- Hosting: [platform]

## Local Setup
[Setup instructions]

## Project Structure
[Key directories and what's in them]

## Key Commands
[Development, test, build, deploy commands]

## Documentation
- ARCHITECTURE.md — System design and schemas
- .clinerules — Development rules for AI
- LEARNINGS.md — Discovered solutions and gotchas
- Tasks/ — Sprint plans and task specs

.clinerules Template

# [Project Name] — Development Rules

## MANDATORY: Read Before Every Task
1. README.md — Current state and setup
2. ARCHITECTURE.md — Schema, routes, components
3. LEARNINGS.md — Known solutions and gotchas
4. The specific task spec for the current task

## Workflow Rules
- Plan mode FIRST. Always. No exceptions.
- One task per conversation.
- Side issues discovered during a task → write task doc, do not fix inline.
- Update LEARNINGS.md when you discover something new.

## Quality Standards (NON-NEGOTIABLE)
- Unit tests for ALL business logic. Minimum 3 tests per function.
- Smoke test after every change.
- Browser test for all UI changes.
- Error handling required. No silent failures.

## Code Standards
- TypeScript everywhere. No `any` types.
- Docstrings on every function.
- Comments on all business logic explaining WHY.
- Section markers (=====) in files over 100 lines.

## Prohibited
- No skipping tests.
- No scope creep — stay in task scope.
- No secrets or API keys in code.
- No console.log in production code.
- No committing broken code.

## Task Completion
Every task must include:
- Confidence score (8/10 minimum to proceed)
- List of completed acceptance criteria
- List of any deferred items (with sprint reference)

ARCHITECTURE.md Template

# Architecture

## System Overview
[Brief description of the system and its main components]

## Tech Stack
[Same as README but with more detail — versions, why chosen]

## Database Schema

### [table_name]
| Column | Type | Constraints | Notes |
|--------|------|-------------|-------|
| id | UUID | PK | |
| created_at | TIMESTAMP | DEFAULT NOW() | |

## Entity Relationships
[How tables relate to each other]

## Route Architecture
[Page/endpoint structure]

## Component Hierarchy
[UI component tree if applicable]

## Auth Flow
[How authentication works end to end]

## Key Technical Decisions
[Decision records: what was chosen, why, what was rejected]

LEARNINGS.md Template

# Learnings

Running log of solutions, discoveries, and gotchas.
Add entries during tasks, not after — while the context is fresh.

---

## [Date] — [Topic]

**Problem:** [What went wrong or was unclear]

**Solution:** [What worked]

**Context:** [Why this matters / when to apply it]

TASK_TEMPLATE.md

# Task X.X: [Task Name]

## Context
Why this task exists and how it fits in the sprint.

## Requirements
1. [Specific, testable requirement]
2. [Specific, testable requirement]

## Technical Approach
- Use [specific file] for [specific purpose]
- Follow patterns in [reference file]
- [Any specific implementation guidance]

## Acceptance Criteria
- [ ] [Testable criterion]
- [ ] Unit tests written and passing (min X tests)
- [ ] Smoke test verified
- [ ] Browser test for UI changes

## Dependencies
- Task X.X must be complete before starting

## Notes
[Any edge cases, gotchas, or additional context]

SPRINT_RULES.md Template

# Sprint Rules

## Sprint Size
- Max 12 tasks per sprint
- Each task should be completable in 1–4 hours
- If a task would generate 300+ lines, split it

## Task Numbering
- Sprint 1 tasks: 1.1, 1.2, 1.3...
- Sprint 2 tasks: 2.1, 2.2, 2.3...

## Statuses
- ⬜ Not started
- 🔄 In progress
- ✅ Complete (8/10+ confidence)
- 🔁 Needs rework

## Task Completion Checklist
- [ ] All acceptance criteria met
- [ ] Tests passing
- [ ] Confidence score 8/10+
- [ ] LEARNINGS.md updated if relevant
- [ ] ARCHITECTURE.md updated if structure changed
- [ ] Code committed

## Discovered Work
If a task reveals additional work needed:
1. Write a task doc for the new work
2. Add it to the sprint backlog
3. Do NOT do it inline in the current task