The Agent Skills Directory

[Indirect Prompt Injection]: The skill facilitates the ingestion of data from Linear issues and comments, which serves as a potential surface for indirect prompt injection attacks.
Ingestion points: Data is retrieved from the description and body fields of Linear entities via the linear_graphql tool.
Boundary markers: There are no explicit instructions for the agent to use delimiters or to disregard instructions embedded within the fetched data.
Capability inventory: The agent possesses the capability to perform mutations in Linear and execute curl commands in the shell.
Sanitization: The instructions do not specify a process for validating or sanitizing the content retrieved from external sources.
[Command Execution]: The skill documentation includes a workflow that involves the execution of curl commands for uploading assets.
Evidence: The section on uploading videos describes the use of curl -X PUT to interact with signed URLs.
Context: This activity is integrated into the primary tool's workflow and includes instructions to use specific signed URLs to maintain security.

linear