The Agent Skills Directory

[COMMAND_EXECUTION]: Uses standard shell utilities (grep, cat, ls) to analyze the local project structure and configuration. These commands are limited to read-only operations and are consistent with the infrastructure auditing purpose.
[CREDENTIALS_UNSAFE]: The skill executes a search for JWT_SECRET in source files (.ts, .js). While intended to verify stateless configuration, this pattern may cause the actual secret value to be loaded into the agent's context and potentially exposed in the audit summary.
[PROMPT_INJECTION]: The skill ingests data from untrusted local files (source code, configuration files, migration scripts) via grep and cat. 1. Ingestion points: Steps 1 through 5 use grep and cat on various project files. 2. Boundary markers: No specific delimiters or instructions are provided to the agent to ignore malicious instructions embedded in the scanned files. 3. Capability inventory: The agent can execute shell commands for file reading and searching. 4. Sanitization: No sanitization or filtering of the file content is performed before processing.

devops-advisor