The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements patterns that ingest data from external sources, creating a potential vector for malicious content to enter the agent's context without sanitization.
Ingestion points: calculate-metadata.md (fetches JSON from a user-provided dataUrl prop), transcribe-captions.md (processes transcription results from user-supplied media files).
Boundary markers: Absent. The instructions do not define delimiters or warnings for the agent when processing this external content.
Capability inventory: The skill uses fetch() for network data, writeFile() for saving transcription results, and npx remotion for video rendering.
Sanitization: Absent. There is no evidence of filtering or validation for the content retrieved from external URLs or media files.
[EXTERNAL_DOWNLOADS]: The skill fetches resources from multiple external services required for video production.
Fetches map imagery and tiles from Google Maps Tiles API and MapTiler.
Retrieves documentation search results from Algolia's API.
Downloads Whisper AI models for audio transcription.
Interacts with ElevenLabs for AI voiceover generation.
[REMOTE_CODE_EXECUTION]: The CesiumFlythrough.tsx component dynamically loads and executes the CesiumJS library from an external CDN (cesium.com) at runtime to enable 3D rendering. While Cesium is a well-known service, this pattern involves executing remote code.
[COMMAND_EXECUTION]: The skill uses the Remotion CLI (npx remotion) for critical developer tasks including starting the preview studio, rendering final video files, and managing package upgrades.
[CREDENTIALS_UNSAFE]: remotion-docs/REFERENCE.md contains a hardcoded Algolia API key (3e42dbd4f895fe93ff5cf40d860c4a85) and Application ID. Although this is a public search-only key for documentation, it is an exposed credential embedded in the instructions.

remotion-best-practices