What problem does it solve? Writing precise, low-false-positive Opengrep rules requires deep knowledge of a codebase's entry points, taint sources, sinks, and sanitizers, which is slow to gather manually. This Skill automates that reconnaissance, producing a structured CODE_RECON.md that feeds directly into the rule-creation stage. ## Core Features & Use Cases - Multi-language codebase mapping: Detects languages, frameworks, and locales across JS/TS, Python, Java/Kotlin, Go, PHP, Swift, and Dart projects using Grep/Glob-first scanning. - Security-focused inventory: Maps HTTP entry points, taint sources, dangerous sinks (SQL, command, template, deserialization, XSS), validators, auth boundaries, crypto usage, hardcoded secrets, and DB schemas. - Auto-scaling depth: Adjusts scan depth (fast/balanced/deep) based on file count, with sampling for very large repositories, and redacts credential values in output. - Use Case: Point it at a large Express.js monorepo to enumerate all routes, raw SQL queries, and missing input validation, then hand the generated CODE_RECON.md to /opengrep-rule-creator to write targeted rules. ## Quick Start Ask the AI to recon the codebase at ./my-project with balanced depth and produce the CODE_RECON.md report for Opengrep rule creation.
