The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it retrieves and processes content from external Miro boards, which may contain malicious instructions.
Ingestion points: Untrusted data is ingested from Miro board items (sticky notes, shapes, text) using the get_items tool described in reference/mcp-tools-reference.md.
Boundary markers: The instructions do not define boundary markers or specific delimiters to separate the skill's system instructions from the content retrieved from external boards.
Capability inventory: The skill has extensive capabilities to modify and delete Miro board content (via update_item and delete_item) and is designed to influence other automated strategy and research workflows.
Sanitization: No sanitization or validation logic is present to filter or neutralize potentially malicious instructions embedded within retrieved board data.
[DATA_EXFILTRATION]: The skill instructs the agent to write telemetry and session data to the local file system.
Evidence: The SKILL.md file contains a mandatory instruction for the agent to write session outcomes, timestamps, and metrics to ~/.claude/skill-analytics/last-outcome-miro.json.
[EXTERNAL_DOWNLOADS]: The skill relies on an external service endpoint for its core functionality.
Evidence: The setup instructions in SKILL.md and reference/setup-guide.md direct the user to connect to the official Miro MCP server at https://mcp.miro.com, which is a well-known and trusted service.

miro-skill