nature-data

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves exclusively as an instructional guide and template framework for constructing academic Data Availability statements. It contains no executable scripts, binaries, tool invocations, or network capability configurations.
  • [INDIRECT_PROMPT_INJECTION]: Assessment of the untrusted data ingestion surface:
  • Ingestion points: Untrusted inputs enter the context when users provide manuscript text, drafts, or data descriptions as described in SKILL.md and static/core/workflow.md.
  • Boundary markers: Absent; there are no instructions to encapsulate user content using delimiters or explicit containment rules.
  • Capability inventory: None; the skill lacks any external tools, shell execution capabilities, file system write operations, or network communication facilities.
  • Sanitization: Absent; input validation or escaping mechanisms are not specified for user-supplied descriptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 04:33 AM
Security Audit — agent-trust-hub — nature-data