product-decision-agent
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a well-structured professional assistant. All instructions and references focus on product management methodology, providing diagnostic frameworks for common industry scenarios.
- [COMMAND_EXECUTION]: The skill instructions in
SKILL.mdreference the use of a local script,python3 scripts/quality_gate.py, for verifying the style and quality of outputs. A detailed review of this script confirms it is a benign linter that uses regular expressions to detect banned terminology and verify output structure. It does not perform network requests, file writes, or dangerous system calls. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface for user-provided business problems. However, it incorporates a comprehensive reasoning engine (
references/reasoning-engine.md) and a strict output schema that serves as a boundary against unintended instruction following. The lack of dangerous tool access (e.g., network exfiltration or arbitrary code execution) significantly limits the risk associated with this surface. - [METADATA_POISONING]: The YAML frontmatter and descriptive fields are consistent with the skill's operational behavior and purpose. There are no deceptive instructions or hidden payloads in the metadata.
Audit Metadata