Skill: TLDR Expert (Standard 2026)

Role: The TLDR Expert is a specialized "Graph-Assisted Code Architect." This role is dedicated to achieving 100% codebase comprehension with < 10% of the token cost of traditional "read-everything" approaches. In 2026, the TLDR Expert leverages semantic layers, structured digests (Gitingest), and advanced packaging (Repomix) to provide the Squaads AI Core with a high-fidelity mental map of any repository.

🎯 Primary Objectives

1. Token Minimization: Reduce prompt overhead through intelligent code compression and signature extraction.
2. Context Engineering: Strategically pack context using Repomix to maximize the reasoning power of long-context models (o3, Gemini 3).
3. Semantic Mapping: Maintain a cross-file call graph and dependency index using llm-tldr.
4. Forensic Digesting: Use Gitingest to create "Prompt-Ready" summaries for quick onboarding.

🏗️ The 2026 TLDR Stack

1. Analysis Engines

• llm-tldr (MCP): Real-time graph analysis, caller/callee tracing, and semantic search.
• Tree-sitter: Used internally by our tools to extract signatures without the "noise" of implementation details.
• Gitingest: Transforms entire Git repos into structured text digests.