finding-podcast-hosts-for-outreach
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [DYNAMIC_EXECUTION]: The input schema includes a
customMapFunctionparameter, which accepts a JavaScript string to transform output data. While this code executes within the external platform's environment, it provides a mechanism for dynamic code execution within the data processing pipeline. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Twitter, such as user biographies and tweet text, to identify and score potential outreach targets. This represents an attack surface where malicious data could influence agent behavior.
- Ingestion points: Twitter profile data and tweets retrieved through the
apidojo~twitter-user-scraperandapidojo~tweet-scrapertools. - Boundary markers: The instructions do not define specific delimiters or guardrails to separate external data from the agent's internal reasoning context.
- Capability inventory: The skill utilizes local shell commands (
node scripts/run_actor.js), network operations viacurland the Apify MCP server, and file system writes for saving results. - Sanitization: There is no evidence of content filtering or sanitization performed on the raw Twitter data before it is evaluated for scoring or output generation.
- [EXTERNAL_DOWNLOADS]: The skill connects to the Apify platform (
api.apify.com) to perform its primary function. These interactions target a well-known service and use actors maintained by the skill's author.
Audit Metadata