The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted content (text, sticky notes, cards) from external Miro boards.
Ingestion points: Untrusted data enters the agent context via the board_list_items, context_explore, and context_get tools in SKILL.md.
Boundary markers: The prompt does not define any delimiters or provide instructions to the agent to disregard commands embedded within the retrieved board data.
Capability inventory: The skill uses tools to explore board structures, retrieve item details, and fetch image or document URLs. It does not appear to possess high-risk capabilities like arbitrary code execution or local file system writes.
Sanitization: There is no evidence of content sanitization, escaping, or validation performed on the data fetched from Miro before it is presented to or processed by the agent.

miro-browse