The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute a shell command that interpolates unsanitized user input. Evidence: The command pnpm -C typescript/cli exec tsx cli.ts warp fork --registry $REGISTRY --warpRouteId $WARP_ROUTE_ID incorporates variables derived from user responses to AskUserQuestion. Risk: Because the agent is not instructed to validate or sanitize these strings, a malicious user could provide input containing shell metacharacters (e.g., ;, &&, or backticks) to execute unauthorized commands on the host system.
[PROMPT_INJECTION]: The skill establishes a surface for indirect prompt injection due to the lack of boundaries between untrusted data and executable instructions. Ingestion points: Warp Route ID and Registry URL gathered via AskUserQuestion. Boundary markers: None present in the prompt instructions to separate the user data from the command context. Capability inventory: Execution of shell commands via pnpm and background shell process management. Sanitization: The skill does not specify any validation, escaping, or filtering for the external content before interpolation.

warp-fork