llm-council-decision-framework
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The installation instructions include
curlcommands to download theSKILL.mdfile from a public GitHub repository (https://raw.githubusercontent.com/aiwithremy/claude-skills-llm-council/main/SKILL.md). This is a standard method for installing the skill itself and targets the author's own repository. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process arbitrary user-provided text for 'council analysis.' Maliciously crafted input could potentially influence the behavior of the simulated advisor personas or the final synthesis, though this is a common risk for analysis-based skills.
- Ingestion points: User input provided following triggers such as 'council this decision' or 'pressure test this.'
- Boundary markers: The skill does not explicitly define markers or 'ignore' instructions for the interpolated user content in its usage examples.
- Capability inventory: The skill primarily functions as a prompting framework and does not define its own file-system writes, network operations, or subprocess executions.
- Sanitization: No explicit sanitization or filtering of the user-provided context is described in the instruction set.
Audit Metadata