The Agent Skills Directory

[SAFE] (SAFE): No malicious patterns detected. The codebase is well-structured and follows industry standards for API wrappers. All remote interactions are limited to the official Apify API domain (api.apify.com).
[COMMAND_EXECUTION] (SAFE): CLI scripts in the skills/ and examples/ directories are intended for local execution via the Bun runtime as part of the tool's normal operation. No unauthorized or dangerous command execution was found.
[EXTERNAL_DOWNLOADS] (SAFE): The skill depends on the legitimate apify-client package and uses standard development dependencies. No suspicious third-party scripts or piped remote executions (e.g., curl|bash) are present.
[INDIRECT_PROMPT_INJECTION] (SAFE): By design, the skill ingests untrusted data from the web (social media, e-commerce). While it lacks explicit prompt delimiters for output, its primary feature is 'code-first' filtering, which significantly reduces the risk of long-context injection by stripping away irrelevant content before it enters the model's reasoning space.
Ingestion points: ApifyDataset.listItems in index.ts, and various transform functions across actor-specific files (e.g., twitter.ts, instagram.ts).
Boundary markers: Absent in returned data strings.
Capability inventory: Apify.callActor in index.ts allows triggering remote scraping tasks.
Sanitization: index.ts provides an optional clean parameter for stripping HTML from scraped items.

Apify