The Agent Skills Directory

[COMMAND_EXECUTION]: The health-check.sh and verify-env.sh scripts execute local CLI tools (vercel, railway, redis-cli, psql) and environment checks. All shell variables and inputs are properly quoted to prevent injection vulnerabilities.\n- [SAFE]: No malicious patterns, such as remote code execution, data exfiltration, or persistence mechanisms, were found. The metadata and implementation are consistent with the skill's stated devops purpose.\n- [PROMPT_INJECTION]: Category 8 (Indirect Prompt Injection) analysis: The skill ingests data from external CLI outputs. The risk of the agent being misled by tool output is mitigated by sanitizing the responses (using head and tr) and wrapping them in a structured JSON schema before returning them to the agent context.

devops-scripts