The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection within its document editing workflow.
Ingestion points: The doc_get tool (referenced in 'Editing Existing Documents') reads content from external Miro board URLs provided by the user.
Boundary markers: The skill instructions do not provide delimiters or "ignore embedded instructions" warnings for the agent when processing the content retrieved from the board.
Capability inventory: The agent has the capability to modify board content via doc_update and doc_create, which could be exploited if malicious instructions are present in the ingested document.
Sanitization: No sanitization or validation of the retrieved markdown content is performed before the agent interpolates it into its context for further actions.

miro-doc