The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The specification defines mechanisms for interpolating external data into sensitive execution contexts, including LLM prompts and database queries.
Ingestion points: The skill shows data being pulled from inputs and tool outputs via placeholders like {input.location}, {input.customer_id}, and {gather.output} in SKILL.md.
Boundary markers: The provided examples do not include delimiters or instructions for the agent to ignore potentially malicious content within these interpolated fields.
Capability inventory: The skill specifies nodes for external API interaction (APINode), direct database access (DatabaseNode), and LLM text generation (LLMNode).
Sanitization: No sanitization or input validation logic is present in the declarative templates shown in the skill.
[EXTERNAL_DOWNLOADS]: The skill recommends installing a library and provides links to external documentation.
Evidence: It includes instructions to pip install pyagentspec and links to Oracle's GitHub repository and technical documentation. These resources are related to the skill's primary purpose and target a well-known organization.

oracle-agent-spec