The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external source (Linear issues, projects, and documents) and possesses write capabilities.
Ingestion points: Data enters the agent context via tools such as get_issue, list_comments, get_document, and search_documentation in SKILL.md.
Boundary markers: The instructions do not specify delimitation or instructions to ignore embedded commands within the ingested data.
Capability inventory: The skill can perform modifications via create_issue, update_issue, create_comment, and create_project across its scripts.
Sanitization: No explicit sanitization or validation of the external content is mentioned.
[EXTERNAL_DOWNLOADS]: The skill references and downloads components from official Linear and project-related services.
Evidence: Fetches configuration from https://mcp.linear.app/mcp and uses npx to retrieve mcp-remote from the npm registry.
[REMOTE_CODE_EXECUTION]: The troubleshooting section suggests running a remote transport client via npx.
Evidence: Instructs the environment to run npx -y mcp-remote https://mcp.linear.app/sse to establish a connection to the Linear MCP server.

linear