The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes biological sequence data from FASTA files and SMILES strings. This ingestion of untrusted data represents a potential surface for indirect instructions, although the scientific context and structured nature of the data minimize this risk. * Ingestion points: FASTA file records (protein, ligand, RNA, DNA) processed in SKILL.md and references/chai_usage.md. * Boundary markers: No explicit delimiters are specified in the instructions to isolate sequence data from the agent's prompt context. * Capability inventory: The skill has access to shell execution (Bash for python and uv) and file system operations (Read, Write, Edit) across all scripts. * Sanitization: The skill relies on the chai_lab library's internal parsing; no explicit sanitization of sequence headers is mentioned.
[EXTERNAL_DOWNLOADS]: The skill initiates the download of model weights and the installation of the chai_lab package from the official Chai Discovery repository. This is a legitimate functional requirement for structural prediction tools.

alterlab-chai