The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The instructions in SKILL.md explicitly direct the agent to retry MCP server installation commands with "escalated permissions" if an initial attempt is blocked by security or permission constraints.
[COMMAND_EXECUTION]: The helper script scripts/fetch-codex-manual.mjs executes curl or curl.exe via execFile, which allows the agent to run system binaries to fetch content.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation and update information from OpenAI developer domains (developers.openai.com). These references are documented here as part of the skill's normal operation to maintain up-to-date documentation.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes external Markdown content from remote URLs which could potentially introduce untrusted instructions.
Ingestion points: Documentation is fetched using scripts/fetch-codex-manual.mjs and scripts/resolve-latest-model-info.js.
Boundary markers: The skill does not implement specific delimiters or warnings to isolate the fetched external content from the model's instruction logic.
Capability inventory: The skill can execute shell commands (curl) and perform file system operations (write to cache).
Sanitization: The documentation fetcher performs SHA-256 hash validation to verify the integrity of the downloaded manual.

openai-docs