The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The scripts scripts/remove_bg.py and scripts/vectorize.py contain logic that attempts to read and parse the user's shell profile (~/.zshrc) to extract API keys if they are not found in the environment. Shell configuration files are highly sensitive and should not be accessed by skill scripts as they often contain SSH keys, database credentials, and other secrets.
[DATA_EXFILTRATION]: Both scripts/remove_bg.py and scripts/vectorize.py use curl to upload local files to external third-party APIs. These scripts take file paths as command-line arguments without any validation. An attacker could exploit this by providing a path to a sensitive file (e.g., ~/.ssh/id_rsa or .env), causing the agent to exfiltrate that file's content to the remote API service.
[COMMAND_EXECUTION]: The skill relies on the subprocess.run method to execute system-level binaries like grep and curl. Using external subprocesses instead of native language libraries for network requests and file operations increases the attack surface and makes it harder to maintain security boundaries.
[EXTERNAL_DOWNLOADS]: The skill interacts with external service domains (api.remove.bg, external.api.recraft.ai) and performs downloads of remote content, such as generated SVG files, which could lead to unauthorized data ingress if the remote source is compromised.

logo-creator