icp-targeting-tiers

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data (company and product descriptions) and interpolates it into a deliverable template (templates/output-template.md). This creates an indirect prompt injection surface where malicious instructions in the input could potentially influence the agent's behavior during the playbook generation.
  • Ingestion points: Business context fields in templates/output-template.md.
  • Boundary markers: The skill does not define clear delimiters or 'ignore embedded instructions' warnings for the ingested data.
  • Capability inventory: A local Python validation script (scripts/check-output.py) that reads files from the filesystem.
  • Sanitization: No sanitization or input validation logic is present to filter potentially malicious instructions from the user-provided data.
  • [COMMAND_EXECUTION]: The skill includes a utility script (scripts/check-output.py) and directs the agent to execute it to verify the quality of the generated playbook. The script is restricted to reading the output file and checking for required terminology, representing a low-risk, functional use of the execution environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 12:47 PM
Security Audit — agent-trust-hub — icp-targeting-tiers