The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data scraped from diverse web sources (e.g., Google Maps, TikTok, Instagram) via Apify Actors, which may contain malicious instructions designed to influence the agent's behavior during summarization.
Ingestion points: Data is retrieved from the Apify API (api.apify.com) and processed by reference/scripts/run_actor.js before being presented to the agent or saved to disk.
Boundary markers: The script implements a "Quick answer" mode that truncates data fields to 100 characters, and the SKILL.md instructions recommend summarizing results rather than repeating them verbatim.
Capability inventory: The agent can execute shell commands to run the provided script, perform network requests to Apify's infrastructure, and write files to the local file system.
Sanitization: The run_actor.js script includes logic to prevent CSV formula injection (CSV Excel Injection) by prefixing potentially executable cells (starting with =, +, @, -) with a single quote.
[DATA_EXFILTRATION]: The skill communicates with the Apify API (api.apify.com) to manage scraping tasks and retrieve results. This network activity is documented and required for the skill's primary function.
[COMMAND_EXECUTION]: The workflow relies on the execution of a local Node.js script to perform the complex tasks of starting, polling, and downloading data from the Apify platform. The script includes regex validation for its arguments to prevent command injection through malformed Actor IDs.

apify-lead-generation