The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and processes data from external YAML configuration files, which is a potential surface for indirect prompt injection.
Ingestion points: The skill reads from configs/targets/targets_dict.yaml, configs/targets/ligand_targets_dict.yaml, and configs/design_tasks/ame_dict_v2.yaml to list or show target definitions.
Boundary markers: The instructions do not specify any delimiters or 'ignore' instructions when reading and displaying file content to the user.
Capability inventory: The skill uses Bash to execute the complexa CLI and other shell utilities (grep, mkdir), and the Write tool to modify the configuration files.
Sanitization: The skill correctly advises the agent to use shell-escaping and YAML quoting for complex inputs like SMILES strings and chain specifications.

complexa-target