The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses authoritative language such as "CRITICAL" and "MUST" to instruct the agent to disregard its internal training data in favor of live documentation from external URLs. While this is intended to ensure code accuracy for a fast-evolving API, the use of high-priority triggers to override internal knowledge is a prompt injection pattern.
[INDIRECT_PROMPT_INJECTION]: The skill establishes a surface for indirect prompt injection by directing the agent to ingest and act upon content from an external source.
Ingestion points: The skill instructs the agent to use WebFetch on multiple subpages of motion.dev within SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between documentation and potential malicious commands within the fetched content.
Capability inventory: The agent uses the ingested data to generate code, component structures, and animation logic.
Sanitization: No sanitization or validation logic is defined for the fetched documentation.
[EXTERNAL_DOWNLOADS]: The skill references official documentation links from motion.dev, which is the primary domain for the library described. These references are used to fetch up-to-date technical guidelines.

motion