AI tools catalog
4 tool cards: use cases, strengths and weaknesses, pricing, alternatives.
Anthropic's terminal AI engineer. It works in a repository, writes code, runs tests, opens PRs, uses MCP servers and slash commands.
Best for: Development in existing repositories, refactoring, debugging, automating developer routine.
Strengths
- Direct access to the file system and shell
- MCP support
- Good code quality
- Headless mode for CI
Weaknesses
- Requires a Pro/Max subscription
- A learning curve for non-developers
Pricing: Included in Claude Pro / Max or the API·Difficulty: Intermediate·Alternatives: codex, cursor, windsurf
OpenAI's AI engineer: a CLI and cloud agents that work in a repository. Opens PRs, runs tasks in a sandbox, integrates with GitHub.
Best for: Long tasks in a sandbox, batch PRs, code review.
Strengths
- Cloud tasks without tying up your local machine
- Good GitHub integration
- Parallel tasks
Weaknesses
- Dependence on the ChatGPT infrastructure
- Less flexible local-environment settings
Pricing: Part of ChatGPT Plus / Pro / Team·Difficulty: Intermediate·Alternatives: claude-code, cursor
A VS Code-based AI IDE with deep model integration. Suited for those who want a familiar editor + a strong AI assistant right inside it.
Best for: Pair programming with the model, refactoring, LLM-level autocomplete.
Strengths
- The familiar VS Code interface
- Support for several models
- Good work with diffs
Weaknesses
- A paid subscription for the best features
- Sometimes overdoes the autocomplete
Pricing: Free with limits; Pro is paid·Difficulty: Intermediate·Alternatives: windsurf, claude-code
An AI IDE focused on agent mode: the model walks the project itself, makes changes, runs commands. An alternative to Cursor.
Best for: Agentic tasks in the IDE, long changes across multiple files.
Strengths
- A strong agent mode (Cascade)
- Transparent agent steps
Weaknesses
- Fewer users and plugins than Cursor
Pricing: Free with limits; paid tiers·Difficulty: Intermediate·Alternatives: cursor, claude-code