The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the use of the sikaru command-line interface for environment management and evaluation tasks. This includes operations for pushing datasets (sikaru environment push), listing environments, and executing evaluation runs (sikaru eval run). These commands are the core functional components of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data provided by the user in JSONL or CSV formats to build evaluation sets.
Ingestion points: External data is ingested through the sikaru environment push command, which reads from files like examples.jsonl or CSV files.
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to help the agent distinguish between data content and its core instructions.
Capability inventory: The skill possesses capabilities to read files, transmit data to the Sikaru platform, and execute evaluation jobs that generate traces and logs.
Sanitization: No explicit sanitization, validation, or escaping mechanisms for the user-provided data are mentioned in the documentation.

sikaru-evals