The Agent Skills Directory

[SAFE]: The skill provides structured templates and documentation for integrating the Prime Intellect Verifiers library, a legitimate tool for building reinforcement learning environments.
[EXTERNAL_DOWNLOADS]: The skill references standard Python packages such as verifiers, datasets, openai, and e2b-desktop. These are well-known libraries from reputable technology organizations and represent expected dependencies for the skill's stated purpose.
[INDIRECT_PROMPT_INJECTION]: The skill describes an architecture where an agent processes data from external datasets and tool outputs. This introduces a potential surface for indirect prompt injection common to agentic systems.
Ingestion points: Data is ingested via datasets.Dataset in create_verifiers_env (SKILL.md) and through tool return values in the rollout loop.
Boundary markers: The provided code templates do not implement explicit delimiters or instructions to ignore embedded content within the tool outputs.
Capability inventory: The generated environments may utilize capabilities such as E2B sandboxes or browser automation as mentioned in the environment archetypes.
Sanitization: While explicit sanitization logic is not provided in the code snippets, the skill documentation notes the importance of truncating tool outputs to prevent context overflow.

generate-verifiers-env