linear-discipline

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is composed entirely of markdown instructions and methodology documents. It does not contain any executable code, scripts, or external dependencies beyond the requirement for an official Linear MCP server.
  • [PROMPT_INJECTION]: The methodology explicitly includes a 'Trust boundary' section in METHODOLOGY.md. This section instructs the agent to treat project content (issues, comments, documents) as untrusted data, specifically warning against following embedded instructions, running commands, or revealing secrets found within project files. This is a proactive defense against indirect prompt injection.
  • [DATA_EXFILTRATION]: No evidence of data exfiltration was found. The skill uses official Linear MCP tools for tracking and explicitly forbids the agent from 'inventing' GraphQL or curl commands to interact with APIs, ensuring data flow remains within authorized channels.
  • [CREDENTIALS_UNSAFE]: The skill refers to AUTH.md for authentication, which correctly directs users to configure their own official Linear MCP credentials through their agent harness (Cursor, Claude Code, etc.) rather than hardcoding or soliciting secrets.
  • [EXTERNAL_DOWNLOADS]: No external downloads or unverifiable dependencies are present. The documentation focuses on using already-installed tools and standard project files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 12:32 AM
Security Audit — agent-trust-hub — linear-discipline