The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill allows the agent to ingest protocol definitions from external CSV and JSON data files, which represents a surface for indirect prompt injection.
Ingestion points: Data is read from external files via the create_protocol_design_from_csv and edit_protocol_design_arms methods as described in SKILL.md and references/protocol-design.md.
Boundary markers: There are no specified delimiters or instructions to the agent to disregard instructions potentially embedded within the data content.
Capability inventory: The skill is capable of performing network operations and modifying project state through the Jinkō SDK.
Sanitization: While explicit sanitization of the data is not mentioned, the skill requires a manual --apply flag or explicit confirmation for any destructive or creative operations, providing a human-in-the-loop safeguard.

jinko-protocol