The Agent Skills Directory

[COMMAND_EXECUTION]: The bash script scripts/openai-docs-mcp.sh executes curl and jq commands to interact with the documentation server. This is a standard and safe implementation for a CLI tool.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation content from https://developers.openai.com/mcp. This is an official documentation endpoint for a well-known and trusted organization.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: External documentation content is fetched via the scripts/openai-docs-mcp.sh script (e.g., in fetch and search subcommands) and surfaced to the agent.
Boundary markers: No explicit delimiters or boundary instructions are used to separate the external data from the agent's internal instructions.
Capability inventory: The skill possesses network access via curl and the ability to execute its own CLI wrapper script.
Sanitization: The script uses jq to parse JSON-RPC responses, which ensures the data is well-formed JSON, but it does not perform semantic sanitization of the markdown documentation content itself.

openai-docs-skill