ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill uses instructional framing such as "Hard Gate", "Do NOT proceed", and "Refusal Conditions". These are used to enforce a specific statistical workflow and do not represent attempts to bypass safety filters or override core system instructions.
- [INDIRECT_PROMPT_INJECTION]: While the skill prompts the user to provide data (hypotheses and metrics) which could technically contain injection patterns, the skill lacks any capabilities (such as network access, file writing, or command execution) that could be exploited by such patterns.
- Ingestion points: User-provided hypotheses, metrics, and assumptions in SKILL.md.
- Boundary markers: None explicitly defined in the skill text.
- Capability inventory: None; no subprocess calls, network operations, or file system access identified.
- Sanitization: None specified.
- [NO_CODE]: The skill consists exclusively of markdown documentation and YAML metadata. It does not include any scripts, binary files, or package dependencies.
- [SAFE]: No evidence of data exfiltration, obfuscation, remote code execution, or persistence mechanisms was found.
Audit Metadata