The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from project-supplied manifest and lock files.
Ingestion points: scripts/dep_scanner.py, scripts/license_checker.py, and scripts/upgrade_planner.py parse manifest files including package.json, requirements.txt, go.mod, Cargo.toml, Gemfile, and their respective lockfiles.
Boundary markers: The skill's output reports do not employ explicit boundary markers or isolation instructions to prevent the agent from misinterpreting metadata (such as package names or versions) as instructions.
Capability inventory: The skill provides vulnerability scanning, license compliance analysis, and upgrade path planning, which are high-influence activities that drive critical development workflows.
Sanitization: The parsing logic extracts literal strings from manifests without verifying the absence of instruction-like patterns that could trigger indirect prompt injection when processed by the agent.

dependency-auditor