patchestry

This skill should be used when the user asks to verify, debug, or generate regression tests for the patchir-decomp structuring pipeline. Operates in two modes: --debug verifies that the default structuring pipeline produces functionally equivalent C output compared to the goto-based baseline (produced via the hidden --emit-flat-baseline flag) and performs JSON ground-truth audit. --test-gen extracts P-Code from a binary via Ghidra headless and generates a LIT test with FileCheck patterns. Trigger the skill when the user asks to "debug patchir output", "analyze decompilation issues", "find structuring bugs", "compare structuring output", "diff structured vs goto", "check structuring quality", "analyze structuring pass", "find structuring issues", "test structuring on JSON", "verify structuring correctness", "test patchir-decomp on JSON", "debug a fixture", "find missing statements", "audit patchir-decomp", "check ClangAST emission", "find emission bugs", "generate test", "create LIT test from binary", "genera

This skill should be used when the user runs "/patchir-review", asks to "review the PR for this repo", "run the repo PR review", "run autonomous PR review", or when .github/workflows/claude-review.yml invokes the skill from CI. It performs a repo-specific review of C/C++, Bash, and Python changes in a pull request, prioritizes correctness and security over style, optionally uses clang-tidy for changed C/C++ files, and posts one inline GitHub comment per confirmed finding.