The Agent Skills Directory

[SAFE]: The skill instructions contain multiple examples of insecure code (e.g., SQL injection, use of eval(), hardcoded secrets). These are explicitly labeled as "BAD" patterns to illustrate vulnerabilities for training and auditing. The skill itself does not execute these dangerous code segments.
[COMMAND_EXECUTION]: The skill utilizes Bash to run a provided Python utility (scripts/security-scan.py) for static analysis and standard security auditing tools such as pip-audit and npm audit. These actions are directly aligned with the skill's purpose of identifying security risks in a codebase.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted source code from external pull requests, which presents a surface for indirect prompt injection if that code contains adversarial instructions meant to mislead the agent.
Ingestion points: The agent reads and analyzes source code files from the local environment using Read, Grep, and Glob tools.
Boundary markers: No specific boundary markers or instructions to ignore instructions embedded in the analyzed code are present.
Capability inventory: The agent has the ability to write to the filesystem and execute shell commands via Bash (Python and NPM).
Sanitization: The security-scan.py script performs static analysis using Python's ast (Abstract Syntax Tree) module, which inspects code structure safely without execution.

code-review-security