What problem does it solve? When working on lci's git analysis features (git-analyze CLI, MCP git_analysis, code_insight git modes, HTTP /git-analyze), developers need to know which source files implement each surface, what invariants and traps exist in the git parsers, and how lci's change analysis compares to tools like CodeScene or code-maat. ## Core Features & Use Cases - Surface-to-code mapping: Traces every entry point (CLI, MCP, HTTP) to its handler and core implementation in src/git/* (Provider, Analyzer, FrequencyAnalyzer, PatternDetector). - Invariants and traps documentation: Records parser byte layouts pinned to git 2.43, ref-safety screening, hunk scoping rules, determinism guarantees, and known dead code such as the unwired FrequencyCache and PatternDetector. - Product comparison: Contrasts lci's hunk-scoped duplicate detection, churn hotspots, and churn x complexity risk matrix against CodeScene, code-maat, git-of-theseus, hercules, and gitinspector, with honest gaps. - Use Case: You are fixing a bug where git_analysis returns wrong findings on a range scope; use this Skill to locate the Analyzer pipeline, the -U0 hunk scoping logic, and the relevant tests before editing. ## Quick Start Use the lci-git-analysis skill to find where the git_hotspots risk matrix is computed and which tests pin its output.
