The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute the codex binary and the ./scripts/run_agent.py orchestration script to route code-focused tasks to AI models.
[DATA_EXFILTRATION]: The skill's health check section suggests running echo $OPENAI_API_KEY, which is a security risk as it exposes sensitive credentials in terminal output and command history logs (data exposure).
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by processing external files with powerful tools. * Ingestion points: Untrusted content from INPUT_FILE enters the agent's context via the run_agent.py script. * Boundary markers: No delimiters or markers are identified in the orchestration examples to isolate input data from instructions. * Capability inventory: The skill has access to the Bash, Read, and Write tools. * Sanitization: There is no documentation regarding the sanitization or validation of data read from input files.

codex-cli-agent