The Agent Skills Directory

[SAFE]: The skill facilitates interaction with the Tamarind Bio platform via its official REST API and MCP server. It explicitly instructs the agent and users on the secure handling of the TAMARIND_API_KEY, recommending the use of environment variables or .env files rather than hardcoding credentials.
[SAFE]: The skill contains proactive security warnings, specifically advising against installing the PyPI package tamarind due to a name collision with an unrelated tool, which mitigates supply-chain confusion risks.
[INDIRECT_PROMPT_INJECTION]: The skill handles user-supplied molecular sequences and structure files which are processed by remote computational tools. While this represents an attack surface, the risk is inherent to the tool's scientific purpose and mitigated by schema validation.
Ingestion points: Molecular sequences (amino-acid strings) and structure files (PDB, SDF, CIF) provided by the user in the settings payload or via file uploads.
Boundary markers: The instructions define chain delimiters (colon :) for multimers and direct the agent to utilize the validateJob endpoint for schema and parameter verification before submission.
Capability inventory: The skill utilizes the requests library to perform network operations against the Tamarind Bio domain and writes result data (ZIP archives) to the local file system.
Sanitization: Relies on platform-side validation and the validateJob API to ensure that inputs conform to expected biological and chemical formats.

tamarind