litmus_agent:
  resources_servers:
    litmus_agent:
      entrypoint: app.py
      domain: knowledge
      verified: false
      description: >-
        Domain-agnostic answer verifier: extracts a model's final answer with an
        answer_format regex and scores it against expected_answer per an
        answer_type taxonomy (float, bool, string). For tool-using rows it can
        also host a sandbox-backed stateful Python code-execution tool. This
        variant backs the tool with the local Apptainer provider (no service or
        API key); swap to litmus_agent.yaml for the OpenSandbox provider.
      value: Reusable scoring for any benchmark whose tasks reduce to "match the expected answer".
      float_rel_tol: ${oc.env:LITMUS_AGENT_FLOAT_REL_TOL,1e-6}
      float_abs_tol: ${oc.env:LITMUS_AGENT_FLOAT_ABS_TOL,1e-6}
      # Sandbox is created lazily on the first tool call, so direct (no-tool)
      # rows never need the apptainer binary present.
      code_exec_tool_name: stateful_python_code_exec
      code_exec_timeout_s: ${oc.env:LITMUS_AGENT_CODE_EXEC_TIMEOUT_S,120}
      code_exec_max_output_chars: ${oc.env:LITMUS_AGENT_CODE_EXEC_MAX_OUTPUT_CHARS,10000}
      code_exec_user: root
      # Local Apptainer provider. Requires the `apptainer` binary on PATH (installed
      # at job start via the NeMo-RL SETUP_COMMAND, see mini_swe_agent README).
      sandbox_provider:
        apptainer:
          exec:
            fakeroot_for_root: true
            concurrency: 32
          create:
            mount_point: /sandbox
            start_timeout_s: 600
            # cgroups-v2 delegation is not guaranteed inside an enroot container;
            # leave resource limits off unless the node supports it. Literal bool
            # (not oc.env) because the apptainer sub-config is a plain dataclass
            # with no string coercion -- "False" would be truthy.
            apply_resource_limits: false
          probe:
            command: printf apptainer-sandbox-ready
            expected_stdout: apptainer-sandbox-ready
            deadline_s: 120
            stable_count: 2
            stable_delay_s: 1.0
      sandbox_spec:
        # Apptainer pulls this on first use; pre-build a local .sif and point here
        # if the compute node has no outbound network.
        image: ${oc.env:LITMUS_AGENT_SANDBOX_IMAGE,docker://python:3.12-slim}
        workdir: /sandbox
        resources:
          cpu: 2
          memory_mib: 4096
        # ttl_s intentionally omitted: Apptainer has no native auto-expiry and
        # ignores it (warns). Lifetime is managed via the per-rollout cleanup.
        metadata:
          harness: litmus_agent
          sandbox-api: apptainer

litmus_agent_agent:
  responses_api_agents:
    simple_agent:
      entrypoint: app.py
      max_steps: 10
      resources_server:
        type: resources_servers
        name: litmus_agent
      model_server:
        type: responses_api_models
        name: policy_model
      datasets:
        - name: example
          type: example
          jsonl_fpath: resources_servers/litmus_agent/data/example.jsonl