conversation-to-linear

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs its stated function using legitimate tools without suspicious behavior. All tool usage (Linear API) is directed toward the author's own project infrastructure.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted conversation history to generate issues. However, it implements several safeguards to mitigate this risk.
  • Ingestion points: Reads and processes all previous messages in the conversation (SKILL.md).
  • Boundary markers: Explicitly instructs the agent to only extract items 'explicitly mentioned', ignore vague context, examples, hypotheticals, and items previously rejected by the user.
  • Capability inventory: Employs the Linear:save_issue tool to perform write operations to an external project management service.
  • Sanitization: Enforces a specific data schema (title, description, priority, labels) and requires the agent to determine if an item is 'concrete enough' before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 06:34 AM