The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a Python script (scripts/probe_app_server.py) that executes the codex command-line utility. It uses subprocess.run and subprocess.Popen to generate JSON schemas, TypeScript bindings, and to initiate a local JSON-RPC session for runtime discovery. These executions are constrained to the specific codex binary and are essential for the skill's primary function of providing 'installed-build truth'.
[EXTERNAL_DOWNLOADS]: The documentation references official external resources from OpenAI, including GitHub repositories (github.com/openai/codex) and the developer portal (developers.openai.com). These references are used for obtaining the most recent official protocol behavior and integration guidelines.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The agent ingests information from local command outputs (via scripts/probe_app_server.py) and external documentation retrieved through the openaiDeveloperDocs MCP.
Boundary markers: The probe script enforces structural boundaries by parsing results as JSON before they are processed by the agent.
Capability inventory: The skill can execute local shell commands (specifically the codex binary) and interact with external documentation APIs.
Sanitization: While JSON parsing provides structural safety for local data, the skill relies on the agent's inherent safety filters when processing unstructured text from official documentation.

codex-app-server