The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions and metadata repeatedly reference the sensitive file path ~/.claude/.env for storing the APIFY_TOKEN. Hidden environment files in user home directories are sensitive as they may contain various platform-specific secrets or configuration data beyond what is required for the skill.
[DATA_EXFILTRATION]: The run_actor.js script transmits the user's APIFY_TOKEN as a query parameter in URLs sent to api.apify.com. Transmitting sensitive tokens in query strings is a risky practice as they can be captured in server logs, proxy logs, or browser history.
[PROMPT_INJECTION]: The skill workflow is designed to ingest and summarize untrusted data from external sources (e-commerce product descriptions and reviews), which constitutes an indirect prompt injection surface.
Ingestion points: The script run_actor.js fetches data from the Apify API (e.g., product details, reviews).
Boundary markers: The instructions do not include specific delimiters or warnings for the agent to ignore instructions embedded in the scraped content.
Capability inventory: The skill possesses network access (fetch) and filesystem write access (writeFileSync).
Sanitization: No sanitization or validation of the external content is performed before it is processed or summarized.
[COMMAND_EXECUTION]: The skill provides several shell command examples that execute a local JavaScript file (run_actor.js) using the node runtime. This script has the capability to write arbitrary data to the local filesystem based on user-supplied parameters.

apify-ecommerce