Skip to content

Latest commit

 

History

History
217 lines (148 loc) · 4.76 KB

cline-workflow.md

File metadata and controls

217 lines (148 loc) · 4.76 KB
title description
The Cline Workflow
Plan → Act → Verify, one task at a time

The Cline Workflow

TLDR

New chat for every task. This prevents context pollution.

Minimal prompting. "Can we plan task 2.3?" is enough—AI reads your docs for context.

Plan before Act. Review the approach before AI starts coding.

Approve commands. You see every terminal command before it runs.

Verify and score. Check the work, assign confidence, close chat, next task.

The Cycle

1. Open new Cline chat
2. "Can we please plan task X.X?"
3. AI reads docs, proposes approach
4. Review plan, ask questions, adjust
5. "Proceed" → Act Mode
6. Approve terminal commands as needed
7. AI completes work, provides confidence score
8. Verify it works
9. Close chat
10. Next task

Every task follows this cycle. No exceptions.

Why New Chat Per Task

Long conversations accumulate garbage:

  • Old debugging tangents
  • Superseded decisions
  • Conflicting context

By message 100, AI is confused.

New chat = fresh start. Your documentation provides continuity, not chat history.

The pattern:

  • Task 1 → New chat → Complete → Close
  • Task 2 → New chat → Complete → Close
  • Task 3 → New chat → Complete → Close

Each task gets AI's full attention without pollution from previous work.

Minimal Prompting

You don't need elaborate prompts. This works:

"Can we please plan task 2.3?"

AI will:

  1. Read your ROADMAP.md to find task 2.3
  2. Read README.md for project context
  3. Check CLAUDE_RULES.md for standards
  4. Look at LEARNINGS.md for gotchas
  5. Propose an approach

If your docs are good, the prompt can be simple.

When to add context:

  • External APIs: "Here's the API docs: [link]"
  • Changed requirements: "Actually, we decided to use X instead of Y"
  • Specific constraints: "This needs to work offline"

Otherwise, trust your docs.

Plan Mode

When you prompt, AI enters Plan Mode:

You: "Can we plan task 2.3 - user authentication?"

AI: "Based on the ROADMAP, this task involves:
- Creating login/register endpoints
- JWT token generation
- Password hashing with bcrypt

Proposed approach:
1. Add User model with email/passwordHash
2. Create auth service for hashing/tokens
3. Add POST /auth/login and /auth/register routes
4. Write tests

Questions:
- Token expiration: 24 hours okay?
- Should I add refresh tokens or keep MVP simple?
- Rate limiting now or defer to Phase 2?"

Answer the questions. Adjust the plan if needed.

Don't skip this. Catching misunderstandings now is cheap. Fixing code later is expensive.

Act Mode

Once the plan is right:

"Looks good. Proceed."

AI switches to Act Mode and starts working:

  • Creates/modifies files
  • Runs commands (with your approval)
  • Writes tests
  • Documents the work

Terminal approval:

Cline wants to run: npm install bcrypt jsonwebtoken
[Approve] [Reject] [Edit]

Quick approvals for standard stuff (npm install, npm test). Careful review for anything destructive (rm, database operations, git push).

Completion

When AI finishes, it should provide:

## Confidence: 8/10

**Done:**
- Login endpoint working
- Register endpoint working
- JWT tokens generating correctly
- Tests passing (8/8)

**Deferred:**
- Rate limiting (Phase 2 per roadmap)
- Refresh tokens (v1.0)

**Notes:**
- Used 24h token expiry as discussed
- Added entry to LEARNINGS about bcrypt cost factor

Your job:

  1. Verify it actually works (run the app, test manually)
  2. Check confidence score makes sense
  3. If below 8/10, fix before moving on
  4. Close chat, open next task

Visual Verification

For UI tasks, ask AI what you should see:

"Before we close, describe what I should see when I test the login flow."

AI: "At localhost:3000/login you should see:
- Email and password fields
- 'Sign In' button
- After correct login: redirect to dashboard
- After wrong password: 'Invalid credentials' error

Try email: test@example.com, password: test123"

Test it. Report any differences. Fix before closing.

When Things Go Wrong

AI seems confused: Your docs might be incomplete or contradictory. Check them.

AI does something unexpected: Stop Act Mode, discuss in Plan Mode, then resume.

Confidence is below 8: Don't move on. Ask AI what's missing and fix it.

Task is taking way too long: It's probably too big. Split it into smaller subtasks.

Quick Reference

Phase What Happens
New chat Fresh context, AI reads docs
Plan Mode AI proposes approach, you review
Act Mode AI executes, you approve commands
Completion Confidence score, verification
Close Done. Next task gets new chat

Next: Task Patterns — How to document completed tasks.