The Agent Skills Directory

[SAFE]: The skill is primarily educational, providing code recipes for the official benchling-sdk. No malicious patterns such as prompt injection, obfuscation, or unauthorized data exfiltration were detected.
[EXTERNAL_DOWNLOADS]: The skill recommends installing the official benchling-sdk via pip. This is a standard and expected dependency for the stated purpose of the skill.
[CREDENTIALS_SAFE]: The documentation explicitly instructs users to store API keys and OAuth secrets in environment variables and warns against committing them to version control. The code examples correctly use os.environ to access these secrets.
[DATA_SURFACE]: The skill describes workflows that ingest external data (FASTA files, spreadsheets). While this represents a potential attack surface for indirect prompt injection if processing untrusted user files, the provided examples are standard developer recipes and do not implement autonomous agent loops that would be vulnerable to such attacks without developer oversight.

benchling-integration