linear-batch-refine
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill manages an indirect prompt injection surface by ingesting data from external Linear issues. This is effectively mitigated through the use of HTML comment delimiters and mandatory schema validation.
- Ingestion points: Issue content is retrieved using Linear MCP tools (
list_issues,get_issue,list_comments). - Boundary markers: AI-managed state is clearly isolated within HTML comments (e.g.,
<!-- linear-ai:plan v1 -->). - Sanitization: All extracted YAML data is validated against JSON schemas using the AJV library in a dedicated script, ensuring the agent only processes well-formed and expected data structures.
- [COMMAND_EXECUTION]: The skill executes a local validation script (
scripts/validate_marked_comments.ts) to verify the structure of Linear issue comments. This script is part of the skill's defensive architecture and uses standard, safe libraries for its tasks.
Audit Metadata