The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The .env file contains a hardcoded API key (ANYSEARCH_API_KEY=as_sk_16c4f23c66e3a3b9c9683f79090fa912).
[DATA_EXFILTRATION]: The batch_search command in all four CLI scripts (anysearch_cli.py, anysearch_cli.js, anysearch_cli.ps1, anysearch_cli.sh) supports reading from local files using the @ prefix (e.g., --queries @/path/to/file). The contents of these files are subsequently sent to the external API at https://api.anysearch.com/mcp, creating a potential path for sensitive data exfiltration.
[COMMAND_EXECUTION]: The README.md and SKILL.md provide detailed instructions for the agent to execute shell commands for platform detection, runtime verification, and file system operations (writing runtime.conf).
[EXTERNAL_DOWNLOADS]: The documentation instructs the agent and user to download the skill package from an external repository (github.com/anysearch-ai/anysearch-skill) using curl or wget.

anysearch