The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill processes untrusted external data from Linear tickets to create test cases.
Ingestion points: The skill accepts ticket titles, descriptions, and comments via $ARGUMENTS (Step 1).
Boundary markers: There are no explicit markers or instructions to treat the ticket content as untrusted data.
Capability inventory: The agent is instructed to write files (Step 7) and execute shell commands such as pnpm test and pnpm build (Step 8).
Sanitization: No sanitization or validation of the ticket input is specified before it is used to generate executable test code.
[Dynamic Execution] (LOW): The skill explicitly directs the agent to generate and run new code at runtime.
Description: Steps 7 and 8 involve creating new test files (.ts or .js) and executing them. Although dynamic code generation is typically a MEDIUM risk, it is the primary purpose of this bug-reproduction framework, justifying a downgrade to LOW severity.
[Command Execution] (SAFE): The skill uses standard development commands like pnpm test, pnpm build, and git log. These are used appropriately within the context of a software development environment.

reproduce-bug