JayKimProject/mirage-code-agent

Mirage Code Agent is a Claude-powered code automation tool that uses Mirage — a unified virtual filesystem that can mount disk, RAM, S3, GitHub, Google Drive and other backends — to explore and modify codebases. It exposes filesystem operations as Claude tools (list_directory, read_file, write_file, search_code, find_files, run_bash) so the model can inspect, edit, and verify projects. Who uses it: developers or automation workflows that need model-driven code changes, test generation, refactors, or sandboxed experiments. Key features: - Mounts targets as DiskResource or RAMResource (persistent or in-memory). - Six filesystem tools for programmatic file operations and shell execution. - Uses anthropic/claude-opus-4-7 with adaptive thinking and prompt caching. - CLI entry (main.py) for tasks like adding type hints, writing tests, refactoring, running linters/tests. - Requires Python 3.12+, anthropic, mirage-ai, python-dotenv. Setup includes cloning, installing deps, and providing ANTHROPIC_API_KEY.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。