Fossil

关于

Fossil is a specialized static analysis toolkit designed to combat the code quality issues prevalent in "vibe-coded" projects, where AI agents generate a significant portion of the codebase. Traditional linters often miss the specific problems that arise from this workflow, such as orphaned functions, structural code duplication, temporary AI scaffolding markers, and uncleaned temporary files. Fossil addresses these challenges by building a comprehensive code graph, allowing it to trace function connectivity and identify issues that simple text search cannot. It functions both as a CLI tool for developers and as an MCP server, empowering AI agents with a deeper understanding of the codebase to improve their output and reduce costly LLM inference rounds.

主要功能

• 1 GitHub stars
• Purpose-built for AI-generated code issues, targeting dead code, clones, scaffolding, and temporary files.
• Builds a cross-file code graph for advanced analysis, enabling AI agents to understand function connectivity and blast radius.
• Functions as an MCP server, allowing seamless integration with AI coding tools like Claude Code, Cursor, and VS Code for real-time analysis.
• Supports static analysis across 15 programming languages with framework-aware detection for popular ecosystems like React, Django, and Spring.
• Written in Rust, providing a fast, single-binary CLI with zero configuration and optional interactive REPL for exploring findings.

使用案例

• Identify and eliminate dead code left behind by AI refactoring or unused helper functions.
• Detect structural code clones and duplicated logic generated by limited AI context windows.
• Clean up AI-generated scaffolding, `TODO` markers, and temporary files that accumulate in repositories.
• Provide AI agents with an intelligent code graph to enhance their understanding of a project's structure before making changes.