report-authoring-specialist
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were detected. The skill follows security best practices by implementing a non-mutating, advisory-only workflow. It explicitly prohibits the execution of search queries (SPL), modification of reports, or the handling of sensitive credentials and raw data.
- [INDIRECT_PROMPT_INJECTION]: The skill manages the attack surface associated with processing user-supplied requirements and observations through rigorous defensive instructions.
- Evidence Chain: 1. Ingestion points: Processes user-provided business outcomes and observations as specified in SKILL.md. 2. Boundary markers: Employs a strict labeling system ([supplied], [documented], [observed], [inferred], [unknown]) and commands the agent to treat inputs as evidence rather than authoritative instructions. 3. Capability inventory: Access is limited to the
webtool for public documentation retrieval; no capabilities for file writing, shell execution, or API mutation exist. 4. Sanitization: Instructions demand that all supplied facts be sanitized and explicitly forbid the ingestion of tokens, credentials, or private provider data.
Audit Metadata