llm-citation-readiness
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external URLs or pasted content to perform a citation readiness audit. This content is then processed by the agent for judgment and scoring, creating a surface where malicious instructions embedded in the target pages could influence the agent's output or behavior.
- Ingestion points: Target URLs and pasted page content defined in the 'Inputs' section of SKILL.md.
- Boundary markers: Absent; there are no explicit instructions to the agent to ignore instructions contained within the fetched content.
- Capability inventory: Execution of local scripts (scripts/citation_score.py, scripts/dfs_client.py, scripts/report_publish.py) and file write operations to the local filesystem.
- Sanitization: Absent; external content is used directly for entity clarity and competitive citability assessments.
- [COMMAND_EXECUTION]: The skill invokes local Python scripts using shell commands with user-provided parameters including keywords, brand names, and URLs. Passing unvalidated user input to command-line arguments is a potential risk factor.
- [EXTERNAL_DOWNLOADS]: The skill utilizes a 'webfetch' tool to retrieve content from external URLs provided by the user and to check 'robots.txt' files during the analysis process.
Audit Metadata