AI Coding Guide

AI-Assisted Software Engineering

Tool Selection

By Richard Osborne, CTO at Visual Hive

Last updated:

TLDR

Start every project in Claude (preferably Opus) inside a Project. The brainstorming and document generation quality is unmatched. This is non-negotiable.

Use Cline or Claude Code for execution. Both work. Cline has built-in Plan/Act modes. Claude Code has plan mode too. Pick what fits your workflow.

Use Claude Projects for ongoing context. Sync your GitHub repo so Claude always knows where the project stands.

Claude Projects: Where It All Begins

Every project starts here. A Claude Project gives you:

  • Persistent file context — upload docs, sync a repo, Claude reads them every conversation
  • Opus-quality brainstorming — the initial conversation quality is dramatically better with Opus. The resulting foundation docs will be 10X better than any other tool or model.
  • Multi-turn discussion — brainstorming should be several turns of back-and-forth, not one prompt and one response

How to use it

  1. Create a Project in Claude
  2. Optionally add initial files (design mockups, API docs, reference projects)
  3. Start a conversation about what you want to build
  4. Have a real discussion — multiple turns, questions, push-back, refinement
  5. Ask Claude to generate your foundation documents
  6. Review, refine, export to your repo

When to come back

  • Adding new features (sync the latest GitHub repo first)
  • Major scope changes
  • Strategic decisions about architecture or direction

Cline: Structured Execution

Cline is a VS Code extension with two explicit modes:

Plan Mode: AI reads your project, proposes an approach, asks clarifying questions.

Act Mode: AI executes the plan, creating/editing files and running commands.

This matches the methodology perfectly:

  1. "Can we plan task 3?" → Plan Mode
  2. Review the plan, adjust if needed
  3. "Proceed" → Act Mode
  4. Approve terminal commands as they come
  5. Verify the result

Why terminal approval matters

Cline wants to run: rm -rf node_modules && npm install
[Approve] [Reject] [Edit]

You see every command before it runs. This prevents disasters.

Setup

  1. Install "Cline" from VS Code extensions
  2. Configure with your API key
  3. Enable extended thinking
  4. Keep autoApproveCommands: false (at least initially)

Claude Code: Terminal-Native Execution

CLI tool for developers who prefer terminal workflows.

$ claude "plan task 3.2 from the sprint plan"

Key features for this methodology:

  • Plan modeclaude --plan or ask it to plan before acting. Critical for fixes and debugging.
  • CLAUDE.md — Claude Code reads this file automatically. Same role as .clinerules for Cline.
  • Project context — reads your repo structure and docs

When to use Claude Code vs Cline

Use Claude Code when you prefer terminal over VS Code, want tighter Git integration, or have quick fixes and small tasks. Use Cline when you want visual Plan/Act mode separation, prefer seeing file changes in the editor, or want explicit terminal command approval for larger, more complex tasks.

Both are valid. The methodology works with either. The important thing is: always plan before acting.

The Tool for Each Phase

Phase Tool Why
Initial brainstorm Claude (Opus, in a Project) Best reasoning, persistent context
Foundation docs Claude (same conversation) Keeps brainstorm context
Task execution Cline or Claude Code Plan → Act → Verify cycle
Fixes & debugging Cline or Claude Code Plan mode first, always
New features Claude (Project with repo synced) Strategic discussion first
Phase audits Claude (fresh conversation) Fresh eyes, no dev context
Quick tweaks Claude Code Fast, low overhead

Extended Thinking: Non-Negotiable

Extended thinking is when AI reasons through a problem before responding. Without it:

"Design a matching algorithm"

AI: "Here's cosine similarity..." [generic solution]

With extended thinking:

AI reasons: "1000 users, needs to be fast, budget constraints, vector DB might be overkill for MVP..."

AI: "For your MVP scale, simple tag overlap will be faster and cheaper. Here's why, and here's when you'd upgrade..."

Extended thinking catches problems before they become expensive.

Cost Expectations

Task Type Tokens Approx. Cost
Simple task 5–10k $0.05–0.15
Medium task 15–25k $0.15–0.30
Complex task 30–50k $0.30–0.60
Brainstorm session 50–100k $1–3

Typical MVP (25–35 tasks + brainstorming): $200–500

The methodology adds ~20% overhead (documentation, scoring, audits). It saves 10x that in avoided rework.

Proof It Works

This methodology built RISE — a desktop Electron app — in 4 weeks for ~$400. And the VH Conference Toolkit — a suite of open-source event tools — using the same process with thorough sprint-based documentation.

Building something with AI?

Talk to Visual Hive →