The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted data from local project files. Ingestion points: The agent is instructed to search through the entire codebase using grep and to list/review API routes, components, and server actions. Boundary markers: There are no explicit instructions or delimiters to prevent the agent from following instructions that might be embedded in the code it is auditing. Capability inventory: The agent executes system commands (grep, git, npm, pip) and generates a deployment report. Sanitization: Content from project files is not sanitized or filtered before analysis.
[COMMAND_EXECUTION]: The skill utilizes standard command-line utilities (grep, git ls-files, npm audit, pip list) to inspect the local development environment and project files for security risks.
[DATA_EXFILTRATION]: The instructions include regex patterns designed to identify hardcoded secrets such as AWS keys, Stripe keys, and GitHub tokens. This data is used for local identification of potential exposure; no external network exfiltration patterns were detected.

bouncer