narrative-opposition-web
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external research corpora, which introduces a surface for indirect prompt injection if the source material contains adversarial instructions.
- Ingestion points: The skill processes a 'research corpus' to populate the Four-Corner Worksheet and warrant tables as described in
SKILL.mdandresources/four-corner-worksheet.md. - Boundary markers: The skill does not define specific delimiters or instructions to treat external data as untrusted content, potentially allowing embedded instructions to influence the agent's output.
- Capability inventory: Analysis of
SKILL.mdand theresources/directory shows the skill is restricted to analytical reasoning and prose generation. It contains no tools for network operations, file writing, or subprocess execution. - Sanitization: There are no documented methods for escaping or validating the content of the research corpus before it is processed by the agent.
Audit Metadata