linear-type-labeler
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Linear issue titles and descriptions to determine classification labels.
- Ingestion points: Data enters the agent context through the
mcp__linear-server__get_issueandmcp__linear-server__query_datatools, which fetch issue content from the Linear workspace. - Boundary markers: The instructions lack explicit delimiters or boundary markers to isolate the fetched issue content from the classification logic.
- Capability inventory: The skill possesses the capability to modify external data via the
mcp__linear-server__save_issuetool. - Sanitization: No specific sanitization or escaping of issue content is defined in the workflow.
- Mitigation: The skill implements a human-in-the-loop checkpoint in Step 5 using
AskUserQuestion, requiring the user to approve the labeling plan before the skill proceeds to modify any issues.
Audit Metadata