run-consulting-research-to-output

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a robust, mandatory pre-research gate called 'Frame Challenge' (references/frame-challenge.md) which requires the agent to explicitly confirm material premises, audience context, and evidence access boundaries before proceeding with any data-intensive tasks.
  • [SAFE]: Access to 'private folders' and 'private content' is explicitly managed via a metadata-only allowlist and requires pre-authorization (SKILL.md), preventing unauthorized data exposure during the research phase.
  • [SAFE]: The skill includes internal quality controls such as a 'derived-number scan' and 'fact-deduplication pass' to ensure numerical integrity and prevent content repetition in final artifacts (references/integration-quality-bar.md).
  • [SAFE]: The skill handles potentially untrusted user briefs through a structured adjudication loop and frame verification, mitigating risks of indirect prompt injection by design (SKILL.md, references/central-case-lead-method.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 10:45 AM
Security Audit — agent-trust-hub — run-consulting-research-to-output