paperclip-runner evalsstatic evidence

← Overview

Agent controls a workspace service

control-workspace-service

The agent stops one active-task workspace service exactly once.

Prompt

Call control_workspace_service with serviceId service-runner-lab and action stop.

Fixture

../fixtures/companies/orientation.json

Authority

{
  "actorId": "actor-engineer",
  "capabilities": [
    "workspace:control"
  ],
  "explicitClaims": [
    "workspace:control"
  ],
  "taskId": "task-orientation"
}

Current results

ModelConfigurationResultEstimated costRun time
claude-sonnet-5live-acpx-0.13.1-claude-acp-0.70.0-sonnet-5PASS$0.037887 estimated2026-09-06T21:21:43.878Z
claude-sonnet-5live-claude-managed-agents-2026-04-01-sonnet-5PASS$0.024077 estimated2026-09-06T21:23:09.184Z
global.anthropic.claude-sonnet-4-6live-aws-agentcore-global-anthropic-claude-sonnet-4-6PASS$0.024972 estimated2026-09-06T21:22:36.986Z
gpt-5.4-minilive-codex-pinnedPASS$0.001236 estimated2026-09-06T21:24:05.824Z
gpt-5.6-sollive-acpx-0.13.1-codex-acp-1.6.2-controlNot run——
openrouter/anthropic/claude-sonnet-5live-opencode-1.18.17-openrouter-claude-sonnet-5PASS$0.024919 estimated2026-09-06T21:24:25.748Z
openrouter/deepseek/deepseek-v4-flash-0731live-opencode-1.18.17-openrouter-deepseek-v4-flash-0731PASS$0.000722 estimated2026-09-06T21:28:53.868Z
openrouter/google/gemini-3.8-flashlive-opencode-1.18.17-openrouter-gemini-3.8-flashPASS$0.020325 estimated2026-09-06T21:25:23.263Z
openrouter/openai/gpt-6-astralive-opencode-1.18.17-openrouter-gpt-6-astraPASS$0.022526 estimated2026-09-06T21:26:18.363Z
openrouter/qwen/qwen3.8-max-0902live-opencode-1.18.17-openrouter-qwen3.8-max-0902PASS$0.015180 estimated2026-09-06T21:27:22.362Z
openrouter/z-ai/glm-5.3live-opencode-1.18.17-openrouter-glm-5.3PASS$0.010921 estimated2026-09-06T21:25:31.965Z

Assertions

control-service
semantic_operation
{
  "id": "control-service",
  "kind": "semantic_operation",
  "maxCalls": 1,
  "minCalls": 1,
  "operationId": "control_workspace_service",
  "requireSuccess": true
}
service-stopped
state_match
{
  "collection": "workspaceServices",
  "id": "service-stopped",
  "kind": "state_match",
  "recordId": "service-runner-lab",
  "where": {
    "status": "stopped"
  }
}
terminal-completed
terminal
{
  "id": "terminal-completed",
  "kind": "terminal",
  "status": "completed"
}
mock-only
network
{
  "id": "mock-only",
  "kind": "network",
  "realPaperclipRequests": 0
}
Full authored test definition
{
  "authority": {
    "actorId": "actor-engineer",
    "capabilities": [
      "workspace:control"
    ],
    "explicitClaims": [
      "workspace:control"
    ],
    "taskId": "task-orientation"
  },
  "checks": [
    {
      "id": "control-service",
      "kind": "semantic_operation",
      "maxCalls": 1,
      "minCalls": 1,
      "operationId": "control_workspace_service",
      "requireSuccess": true
    },
    {
      "collection": "workspaceServices",
      "id": "service-stopped",
      "kind": "state_match",
      "recordId": "service-runner-lab",
      "where": {
        "status": "stopped"
      }
    },
    {
      "id": "terminal-completed",
      "kind": "terminal",
      "status": "completed"
    },
    {
      "id": "mock-only",
      "kind": "network",
      "realPaperclipRequests": 0
    }
  ],
  "description": "The agent stops one active-task workspace service exactly once.",
  "fixture": "../fixtures/companies/orientation.json",
  "id": "control-workspace-service",
  "prompt": "Call control_workspace_service with serviceId service-runner-lab and action stop.",
  "schema": "paperclip-runner/eval-case/v1",
  "title": "Agent controls a workspace service"
}