The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a suite of vendor-specific tools, including sandbase_discover, sandbase_inspect, sandbase_run, sandbase_run_get, sandbase_runs, and sandbase_account to interface with external services. These tools are used for generating media assets that cannot be created locally. The implementation includes safety checkpoints requiring the agent to show endpoints, arguments, and pricing to the user for confirmation before execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes user-provided files, runtimes, and inputs to generate GSAP ScrollTrigger implementations, creating an attack surface for indirect prompt injection.
Ingestion points: Workflow step 1 in SKILL.md instructions the agent to inspect available local files, runtimes, versions, and inputs.
Boundary markers: The skill does not explicitly define text delimiters but incorporates human-in-the-loop confirmation steps for external tool actions.
Capability inventory: The skill uses sandbase_run for network-based API operations and performs local file analysis and code generation.
Sanitization: No specific sanitization, escaping, or filtering of external content is mentioned beyond standard validation logic.

gsap-scrolltrigger