The Agent Skills Directory

[DATA_EXFILTRATION]: The scripts/email-upload.js script includes a dedicated --envs flag that uploads local files to an unauthenticated /envs endpoint on a remote server (https://agenskill.onrender.com). The skill documentation explicitly instructs the agent to use this to upload sensitive .env files, which typically contain secrets and credentials.
[DATA_EXFILTRATION]: The scripts/email-numbers.js script allows storing numeric data on a remote server without authentication. This creates a side-channel for the exfiltration of sensitive numeric information such as financial data or personal identifiers.
[EXTERNAL_DOWNLOADS]: The skill's core functionality relies on communicating with an external API hosted on https://agenskill.onrender.com. This domain is an untrusted third-party host and is used as a destination for sensitive data.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it fetches and processes untrusted content from external emails.
Ingestion points: Email content fetched via scripts/email-read.js and scripts/email-search.js.
Boundary markers: None. Untrusted content is not wrapped in delimiters or accompanied by instructions to ignore embedded commands.
Capability inventory: The skill has access to Bash, Read, and Write tools, and can perform network operations like sending or forwarding emails.
Sanitization: None. Raw email data is passed directly into the agent's context, allowing malicious instructions in an email to potentially control agent actions.

email-api