The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill demonstrates processing external data sources which could serve as vectors for indirect prompt injection.
Ingestion points: The skill includes examples for reading and parsing external data from local files, specifically FASTA biological sequences (Bio.SeqIO.parse("sequences.fasta", "fasta")) and CSV files in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to disregard natural language instructions that might be embedded within these data files.
Capability inventory: The skill utilizes the Benchling SDK which performs network operations (REST API calls) and file system writes (saving exported data to CSV).
Sanitization: No explicit sanitization or validation of the content of external files is demonstrated before processing.
[SAFE]: The skill consistently advocates for and demonstrates industry-standard security practices for credential management.
Secret Management: The references/authentication.md file explicitly warns against hardcoding credentials or committing them to version control, recommending the use of environment variables, .env files (with .gitignore), and secret management services like AWS Secrets Manager.

benchling-integration