shell-command-verification-robustness

What problem does it solve? Gates that grep source files, docs, or workflows for command invocations silently miss wrapped invocations (sudo, env, variable assignments, flags with separate values) or fire on ordinary prose like "make sure the build passes", producing green gates that verify nothing or false reds on valid documents. ## Core Features & Use Cases - Failure-direction analysis: Decide whether a miss means silent coverage drift or a false red, then bias the matcher toward the loud failure. - Invocation-shape handling: Split lines on shell separators and strip leading assignments, wrappers (sudo, env, time, xargs), shell keywords, and option flags whose value is a separate word (e.g. make -C dir lint). - English-verb guards: Strip comments before matching and require a real separator ahead of the command name so prose like # make sure the build passes never matches. - Use Case: Writing a Bats make-target coverage contract or a TypeScript doc-references linter that verifies every documented make <target> actually exists, with positive and negative fixtures pinning the matcher's behavior. ## Quick Start Ask the AI to write a grep-based gate that verifies every make target referenced in the documentation actually exists, with fixtures covering wrapped invocations and prose negatives.