SHEET 32 · AI Tooling
Codebase Analysis Suite
9 parallel Claude subagents + tree-sitter AST parsing, a custom DSL for surviving context limits, and four-path effort estimation
no public build · runs locally
2025localAI Tooling
About
A Python system that analyzes any external codebase by combining 9 specialized Claude Opus subagents (up to 8 in parallel) with static analysis tools (Radon, Lizard, tree-sitter AST parsing). Produces structured intermediate documents in a custom DSL format designed to survive context limits, then compiles them into interactive HTML reports with Chart.js charts, Mermaid architecture diagrams, and four-path effort estimation (Agency/Solo x Conventional/AI-Assisted with complexity multipliers).
9 FEATURES
Key features
- 01Analyzes any external codebase from a central installation without modifying the target
- 029 specialized Claude subagents: module, architecture, data model, dependency, test, docs, effort, report, orchestrator
- 03Custom DSL intermediate format — human-readable and machine-parseable, survives context limits
- 04Deep analysis pipeline: tree-sitter AST parsing, business logic classification, coupling metrics, hotspot detection
- 05Four-path effort estimation: Agency/Solo x Conventional/AI-Assisted with complexity multipliers and cost ranges
- 06Publication-quality HTML reports with dark/light mode, sortable tables, radar charts, and Mermaid diagrams
- 07Stack detection (React/Next.js, Node.js, fullstack) with framework-specific pattern recognition
- 08Parallel processing with wave-based execution and dependency barriers
- 09Multi-language complexity analysis: Python, JS/TS, Java, Go, Rust, Ruby, PHP, C/C++, C#, Swift, Kotlin
1 NEARBY SHEETS