xy-kickoff
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run a bundled Python script (
atoms-search.py) to perform keyword-based searches within a specialized knowledge base of business cases. - [EXTERNAL_DOWNLOADS]: The search tool communicates with
https://api.xyskill.xyzto fetch data. This domain is associated with the vendor's infrastructure for the 'XY' skill suite and is used for core functionality. - [INDIRECT_PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it processes user descriptions of business goals and procrastination behaviors.
- Ingestion points: Phase 1 involves a direct prompt for the user to describe their current situation and obstacles.
- Boundary markers: The agent is given a specific classification framework (six diagnostic signals) to process user input, which helps maintain operational boundaries.
- Capability inventory: The skill allows the agent to read and write a local session log (
~/.xy/profile.md) and execute the local search script. - Sanitization: The instructions guide the agent to transform user input into short, 2-4 character keywords for the search script, providing a layer of sanitization against direct injection through tool arguments.
Audit Metadata