remotion-best-practices

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The required workflow “Looking up Remotion APIs and documentation” uses Algolia search and then fetches Remotion documentation pages as Markdown chosen from search results, so outsider-authored free text from search results/pages can be ingested at runtime via fetched .md content.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The remotion-docs skill explicitly instructs the agent to perform runtime fetches (Algolia POST to https://plsduol1ca-dsn.algolia.net/1/indexes/*/queries?x-algolia-api-key=3e42dbd4f895fe93ff5cf40d860c4a85&x-algolia-application-id=PLSDUOL1CA and fetching pages like https://www.remotion.dev/docs/use-video-config.md, https://www.remotion.dev/docs/sequence.md, https://www.remotion.dev/docs/lambda/rendermediaonlambda.md) and to use that retrieved Markdown as the current documentation to implement behavior, which is runtime fetching of external content that directly drives the agent's instructions.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged the Algolia API key because a literal, high-entropy token appears inline in the example POST URL: "3e42dbd4f895fe93ff5cf40d860c4a85" — it is a long, random-looking string and therefore meets the definition of a secret (an API key) present in the documentation.

Ignored items / not flagged:

• Environment variable names such as REMOTION_MAPTILER_KEY, REMOTION_GOOGLE_MAPS_API_KEY, REMOTION_MAPBOX_TOKEN, and other references to GOOGLE_KEY are only variable names (placeholders) and do not contain literal credential values.
• Guidance URLs, example commands, and simple example strings (e.g., "your-service-role-key" style placeholders) are documentation placeholders and intentionally ignored.