✻ Effectively using Claude Code in July 2025
Thushan Fernando
✻ Welcome to Claude Code
██████╗██╗ █████╗ ██╗ ██╗██████╗ ███████╗ ██╔════╝██║ ██╔══██╗██║ ██║██╔══██╗██╔════╝ ██║ ██║ ███████║██║ ██║██║ ██║█████╗ ██║ ██║ ██╔══██║██║ ██║██║ ██║██╔══╝ ╚██████╗███████╗██║ ██║╚██████╔╝██████╔╝███████╗ ╚═════╝╚══════╝╚═╝ ╚═╝ ╚═════╝ ╚═════╝ ╚══════╝ ██████╗ ██████╗ ██████╗ ███████╗ ██╔════╝██╔═══██╗██╔══██╗██╔════╝ ██║ ██║ ██║██║ ██║█████╗ ██║ ██║ ██║██║ ██║██╔══╝ ╚██████╗╚██████╔╝██████╔╝███████╗ ╚═════╝ ╚═════╝ ╚═════╝ ╚══════╝
Press ENTER to continue
✻ Welcome to Claude Code!
/help for help, /status for your current setup
cwd: /home/thushan/projects/talk-claude-code/
Tips for getting started:
Pasting a prompt into the terminal can be a bit awkward.
Fastest and most cost-effective. Great for lightweight coding tasks, scripts and automation flows.
Balanced power and speed. Ideal for everyday dev tasks, debugging and broader code edits.
Most capable. Best for complex reasoning, deep refactors and multi-step agentic workflows.
Use your existing Anthropic Claude monthly or prepaid yearly plan with Claude Code.
Pro / Max (5x | 20x) Levels - see plans
For periods of heavy, short-term demand, such as codebase migrations, massive refactoring, hackathons or crunch time.
| Model | Input | Output | Context | Best For |
|---|---|---|---|---|
|
🚀
Haiku 3.5
|
$0.80 / 1M | $4 / 1M | 200K | Quick edits, simple tasks |
|
🔥
Sonnet 4
|
$3 / 1M | $15 / 1M | 200K | Daily development |
|
🧠
Opus 4
|
$15 / 1M | $75 / 1M | 200K | Complex refactors |
| 💡 Prices as of June 2025. 1M = 1 million tokens | ||||
With a Subscription, you'll have limits......and Claude will tell you!
You'll auto-switch from Opus to Sonnet when you do!
Make sure you have Node v18+
✻ Welcome to Claude Code
Let's get started.
Choose the text style that looks the best with your terminal:
...✻ Welcome to Claude Code
Claude Code can now be used with your Claude subscription or billed based on API usage through your Console account.
Select login method:
Avoids the 'don't code, just asking' messages.
No unwanted edits, consistent formatted response & efficient token use
TIP: Plan into an md file, iterate then execute in Auto mode.
Claude.md is a succinct project-specific instruction file.
Bit like .cursor/rules/ or .junie/guidelines.md
# Project Overview This is a .NET 8 C# web application using ASP.NET Core, Entity Framework Core, and Razor Pages. It’s a multi-layered architecture with Models, Services, and Controllers. ## Tech Stack - .NET 8 - ASP.NET Core - C# - Entity Framework Core - SQL Server - xUnit for tests ## Project Structure - `Controllers/` → API endpoints (REST) - `Models/` → Domain models and DTOs - `Services/` → Business logic - `Data/` → DBContext and EF configurations - `Pages/` or `Views/` → Razor Pages or MVC Views - `wwwroot/` → Static assets - `Tests/` → Unit tests (xUnit) ## Commands - `dotnet build` - `dotnet run` - `dotnet test` - `dotnet ef migrations add` - `dotnet ef database update` ## Code Style - Use PascalCase for public types and methods - Use camelCase for private fields - Async methods end in `Async` - Use expression-bodied members where clear - Prefer dependency injection via constructor ## Repository Guidelines - Branches: `feature/ABC-123-name`, `bugfix/ABC-456-name` - Commits: ` : short summary` (e.g., `fix: handle null in service`) - Prefer squash and merge ## Important Files - `Program.cs`, `Startup.cs` → App entry/config - `appsettings.json` → Configuration - `Data/ApplicationDbContext.cs` → DB context - `Services/*.cs` → Core business logic ## Do Not Touch - Legacy integrations in `Services/Legacy/` - `Startup.cs` bootstrapping unless explicitly asked - Auth setup in `Program.cs` unless requested - Do not generate UI markup unless requested
# Project Overview Python 3.11 project for data processing + API backend using FastAPI and SQLAlchemy. Structured for clarity, testability, and CLI usage. ## Tech Stack - Python 3.11 - FastAPI - SQLAlchemy - Alembic (migrations) - Pytest - Pydantic (schema validation) ## Project Structure - `app/` → Main app code - `api/` → Route handlers - `models/` → DB models (SQLAlchemy) - `schemas/` → Pydantic models - `services/` → Business logic - `core/` → Settings, logging, startup - `tests/` → Pytest-based unit/integration tests - `migrations/` → Alembic migration scripts - `scripts/` → One-off utilities ## Commands - `uvicorn app.main:app --reload` - `pytest` - `alembic upgrade head` - `black .` (format) - `mypy .` (type checks) ## Code Style - Use `black`, `ruff`, `mypy` - Type hint everything - Prefer f-strings - One class/function per file where possible - Avoid global state ## Repo Etiquette - Branches: `feature/ABC-123-name` - Commits: `feat: short summary`, `fix: ...`, `refactor: ...` - Prefer PRs with linked issues ## Core Files - `app/main.py` → FastAPI app - `app/core/settings.py` → Env config - `app/models/base.py` → Declarative base - `app/services/*.py` → App logic - `scripts/*.py` → CLI tools ## Do Not Touch - `.env` and `settings.py` config handling - Alembic-generated scripts - Migration history in `migrations/` - Legacy adapter modules in `app/integrations/`
| Location | Scope | Use Case | VCS Safe? |
|---|---|---|---|
~/.claude/CLAUDE.md |
Global | Personal defaults, universal behaviours | No (local-only) |
your-repo/CLAUDE.md |
Project-wide | Team-wide shared context | ✅ Yes |
your-repo/feature/CLAUDE.md |
Scoped | Fine-grained control per folder/feature | ✅ Yes |
CLAUDE.local.md |
Local override | Personal tweaks not checked into git | No (add to .gitignore) |
| 💡 Files are merged in order: local → subdir → project → global | |||
Claude.mdI'll analyze the codebase to create a comprehensive CLAUDE.md file. Let me start by exploring the project structure and key files.
Update Todos
✻ Herding… (220s · ⚒ 1.4k tokens · esc to interrupt)
docs/ & docs/_promptsClaude.mddocs/ for contextLittle less conversation,
& lot more action!
You can encourage Claude to think harder with certain keywords in prompt.
ultrathink is ideal for Opus model usage
ultrathink apparently allocates ~32K tokens
Claude gives more consideration to planning, design & debugging
Claude doesn't remember past sessions...
...it builds short-term memory as you work.
Use /compact to summarise your session and free up context.
Helps before long tasks like refactors or tests code/fixes
Doesn't persist anything — just condenses runtime memory
Claude Code does manage it's own memory...
...but you can help it out.
Use with caution - no confirmation, no plan, just do it, Claude.
Development environments have been destroyed in this mode!
Jean Claude van Daaaayeeeem!
╭──────────────────────────────────────────╮ │ │ │ Claude Code Token Usage Report - Daily │ │ │ ╰──────────────────────────────────────────╯ ┌────────────┬────────────────────┬───────────┬───────────┬───────────────┬──────────────┬───────────────┬─────────────┐ │ Date │ Models │ Input │ Output │ Cache Create │ Cache Read │ Total Tokens │ Cost (USD) │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-14 │ - opus-4 │ 1,225 │ 27,748 │ 897,182 │ 29,576,651 │ 30,502,806 │ $59.63 │ │ │ - sonnet-4 │ │ │ │ │ │ │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-15 │ - opus-4 │ 166 │ 5,740 │ 85,133 │ 1,508,838 │ 1,599,877 │ $4.29 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-16 │ - opus-4 │ 429 │ 10,594 │ 291,712 │ 9,421,623 │ 9,724,358 │ $20.40 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-17 │ - opus-4 │ 874 │ 67,454 │ 1,020,404 │ 22,927,970 │ 24,016,702 │ $58.60 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-18 │ - opus-4 │ 3,088 │ 136,700 │ 2,581,984 │ 65,141,729 │ 67,863,501 │ $156.42 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-19 │ - opus-4 │ 3,543 │ 164,705 │ 2,547,465 │ 54,788,136 │ 57,503,849 │ $142.35 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-20 │ - opus-4 │ 1,566 │ 36,905 │ 1,147,892 │ 24,487,830 │ 25,674,193 │ $61.05 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-21 │ - opus-4 │ 323 │ 13,887 │ 760,264 │ 5,514,827 │ 6,289,301 │ $23.57 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ 2025-07-22 │ - opus-4 │ 718 │ 16,185 │ 1,868,118 │ 12,644,390 │ 14,529,411 │ $55.22 │ ├────────────┼────────────────────┼───────────┼───────────┼───────────────┼──────────────┼───────────────┼─────────────┤ │ Total │ │ 11,932 │ 479,918 │ 11,200,154 │ 226,011,994 │ 237,703,998 │ $581.54 │ └────────────┴────────────────────┴───────────┴───────────┴───────────────┴──────────────┴───────────────┴─────────────┘
Next time, we'll get our hands dirty with Claude Code
...but hopefully you can use Claude Code in the meantime
Let's discuss!