warehouse-flow

Fail

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: HIGHDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill accesses and writes to sensitive directory paths in the user's home folder.
  • Evidence: The 'SELF-EVOLUTION TELEMETRY' section of the skill instructs the agent to search for and write metadata to the ~/.claude/projects/ directory, which contains platform-specific project information.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the ingestion of untrusted warehouse configuration and operational data.
  • Ingestion points: Reads warehouse topology, location master data, SKU profiles, and WMS integration data from files in the project environment.
  • Boundary markers: Absent. The instructions do not define delimiters or provide warnings to ignore instructions that might be embedded in the processed warehouse data.
  • Capability inventory: The skill has the ability to write files to the local filesystem (e.g., docs/warehouse-flow-analysis.md) and log data to sensitive directories in the home folder.
  • Sanitization: Absent. There is no mention of sanitizing, validating, or escaping the data ingested from the external configuration files.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 23, 2026, 10:58 AM
Security Audit — agent-trust-hub — warehouse-flow