The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to retrieve and analyze sensitive personal data from the user's iMessage database.\n
Evidence: It utilizes tools to access contact names, message frequency, shared links, and the full content of message bodies (body_text).\n
Context: Accessing private communication history is a high-sensitivity operation, although it aligns with the skill's stated purpose.\n- [COMMAND_EXECUTION]: The skill provides a tool that allows the agent to execute arbitrary logic against a database.\n
Evidence: The query tool enables the execution of arbitrary SQL statements, giving the agent direct control over how it interacts with the underlying message data storage.\n- [PROMPT_INJECTION]: The skill exposes the agent to indirect prompt injection from untrusted message content.\n
Ingestion points: Data from the messages table, specifically the body_text and text fields, is ingested into the agent's context during queries.\n
Boundary markers: No specific boundary markers or instructions are provided to help the agent distinguish between its instructions and the data being analyzed.\n
Capability inventory: The agent has the ability to perform arbitrary SQL queries and summarize message content using the tools provided by the imessage-analysis server.\n
Sanitization: The skill lacks any sanitization or validation steps to neutralize potential instructions embedded within the user's message history.

query-messages