tldr

The tldr skill provides engineering-grade codebase analysis designed to minimize token consumption during LLM interactions. By generating LLM-ready summaries of functions, call hierarchies, and architectural dependencies, it allows agents to navigate large repositories without the overhead of processing raw source files. It is specifically built for complex refactoring tasks, debugging sessions, and architectural exploration where context window management is critical for success.

• Advanced semantic search using 5-layer embeddings (AST, call graph, CFG, DFG, PDG) to find code by meaning rather than exact strings.

• Automated impact analysis to identify all potential callers and affected tests, essential for safe refactoring.

• Program slicing for debugging (backward and forward) to determine how variables or data flows influence specific lines of code.

• Token-efficient function context extraction that provides high-level summaries instead of full file content.

• Comprehensive diagnostic tools including dead code detection, control flow graphs, data flow graphs, and project structure visualization.

• Support for 17 programming languages with automatic detection, ensuring compatibility across diverse tech stacks.

• Always utilize tldr context as a primary step before reading large files to save context window space.

• Use the impact tool before performing any refactoring to ensure downstream dependencies are accounted for.

• Leverage the semantic search for natural language queries like 'how does authentication work' or 'find error handling patterns'.

• Integrate with git diffs via change_impact to identify exactly which tests need running after modifications.

• This tool is designed for performance; it utilizes a background daemon to keep indexes warm and queries fast.

• While it provides deep analytical power, use standard grep-like tools only for simple, exact string matching.