xy-publish-guard

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local utility script (scripts/atoms-search.py) to search for specific risk patterns and regulatory requirements.
  • [EXTERNAL_DOWNLOADS]: The search script connects to https://api.xyskill.xyz to retrieve data from the 'XY' atom library. This network operation is used to maintain up-to-date analysis criteria.
  • [DATA_EXFILTRATION]: Keywords derived from user-provided content are transmitted to the external domain https://api.xyskill.xyz to facilitate the search for relevant risk cases. This behavior is documented as part of the skill's search mechanism.
  • [INDIRECT_PROMPT_INJECTION]: As the skill is designed to ingest and analyze untrusted user-provided marketing copy and scripts, it is exposed to potential prompt injection via processed data. The skill mitigates this through a prescriptive multi-phase execution flow and a mandatory report structure that prevents the agent from following embedded instructions.
  • [SAFE]: The local search script attempts to read the application-specific configuration file ~/.xy/config.json and scans standard skill installation paths (e.g., ~/.claude/skills/) to find its data assets. These file operations are restricted to finding the skill's own resources.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 05:22 AM
Security Audit — agent-trust-hub — xy-publish-guard