skills/yogsoth-ai/de-anthropocentric-research-engine/establish-empirical-baseline/Gen Agent Trust Hub
establish-empirical-baseline
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted performance data from external sources, presenting a vulnerability surface where instructions hidden in scientific data could influence agent behavior.
- Ingestion points: External research papers, leaderboards, and inputs such as method_records and benchmark_or_task defined in SKILL.md.
- Boundary markers: Absent; the skill does not establish explicit delimiters or instructions to ignore embedded commands within the processed data.
- Capability inventory: No dangerous capabilities such as arbitrary command execution, network exfiltration, or unauthorized file system writes were detected in the orchestration instructions.
- Sanitization: Absent; although the protocol includes a quality audit step, there is no explicit sanitization or validation of raw external data before it enters the context.
Audit Metadata