The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx @useclawlink/cli to download and run the ClawLink CLI tool. This package belongs to the skill's authoring organization and is used for its primary functionality.
[COMMAND_EXECUTION]: The instructions direct the AI agent to execute shell commands via npx to perform Trello operations such as reading boards, creating cards, and managing checklists.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text to determine Trello actions. While this is a potential ingestion point for indirect instructions, the skill specifically instructs the agent to perform read operations first and require user confirmation before any write actions, which is a significant mitigation.
Ingestion points: User input provided to npx @useclawlink/cli actions trello in SKILL.md.
Boundary markers: The skill contains explicit prose instructions: "Reads first; confirm with the user before any write."
Capability inventory: The skill uses subprocess execution via the npx command to interact with the Trello API.
Sanitization: Content sanitization is handled by the vendor's CLI tool.

trello