paperclip-runner evalsstatic evidence

← Overview

Agent reads context before changing task state

context-before-action

The agent must orient itself before recording a progress mutation on its assigned mock task.

Prompt

Please begin work on your assigned task. Record a brief progress update describing the immediate next step you will take.

Fixture

../fixtures/companies/orientation.json

Authority

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

Current results

ModelConfigurationResultEstimated costRun time
claude-sonnet-5live-acpx-0.13.1-claude-acp-0.70.0-sonnet-5FAIL$0.150857 estimated2026-09-06T21:22:17.256Z
claude-sonnet-5live-claude-managed-agents-2026-04-01-sonnet-5FAIL$0.030822 estimated2026-09-06T21:22:17.068Z
global.anthropic.claude-sonnet-4-6live-aws-agentcore-global-anthropic-claude-sonnet-4-6PASS$0.042075 estimated2026-09-06T21:22:17.330Z
gpt-5.4-minilive-codex-pinnedPASS$0.001289 estimated2026-09-06T21:24:05.002Z
gpt-5.6-sollive-acpx-0.13.1-codex-acp-1.6.2-controlPASS$0.007670 estimated2026-09-06T21:22:16.918Z
openrouter/anthropic/claude-sonnet-5live-opencode-1.18.17-openrouter-claude-sonnet-5FAIL$0.034968 estimated2026-09-06T21:24:04.307Z
openrouter/deepseek/deepseek-v4-flash-0731live-opencode-1.18.17-openrouter-deepseek-v4-flash-0731FAIL$0.002742 estimated2026-09-06T21:28:17.322Z
openrouter/google/gemini-3.8-flashlive-opencode-1.18.17-openrouter-gemini-3.8-flashPASS$0.018478 estimated2026-09-06T21:24:46.942Z
openrouter/openai/gpt-6-astralive-opencode-1.18.17-openrouter-gpt-6-astraFAIL$0.069529 estimated2026-09-06T21:26:07.863Z
openrouter/qwen/qwen3.8-max-0902live-opencode-1.18.17-openrouter-qwen3.8-max-0902FAIL$0.025202 estimated2026-09-06T21:27:20.110Z
openrouter/z-ai/glm-5.3live-opencode-1.18.17-openrouter-glm-5.3FAIL$0.035174 estimated2026-09-06T21:25:27.738Z

Assertions

context-call
semantic_operation
{
  "id": "context-call",
  "kind": "semantic_operation",
  "minCalls": 1,
  "operationId": "get_task_context",
  "requireSuccess": true
}
progress-call
semantic_operation
{
  "id": "progress-call",
  "kind": "semantic_operation",
  "maxCalls": 1,
  "minCalls": 1,
  "operationId": "report_progress",
  "requireSuccess": true
}
context-before-progress
operation_order
{
  "after": "report_progress",
  "before": "get_task_context",
  "id": "context-before-progress",
  "kind": "operation_order"
}
one-progress-comment
state_effect
{
  "collection": "comments",
  "id": "one-progress-comment",
  "kind": "state_effect",
  "maximumAdded": 1,
  "minimumAdded": 1
}
terminal-completed
terminal
{
  "id": "terminal-completed",
  "kind": "terminal",
  "status": "completed"
}
transcript-complete
transcript_complete
{
  "id": "transcript-complete",
  "kind": "transcript_complete"
}
mock-only
network
{
  "id": "mock-only",
  "kind": "network",
  "realPaperclipRequests": 0
}
Full authored test definition
{
  "authority": {
    "actorId": "actor-engineer",
    "capabilities": [],
    "explicitClaims": [],
    "taskId": "task-orientation"
  },
  "checks": [
    {
      "id": "context-call",
      "kind": "semantic_operation",
      "minCalls": 1,
      "operationId": "get_task_context",
      "requireSuccess": true
    },
    {
      "id": "progress-call",
      "kind": "semantic_operation",
      "maxCalls": 1,
      "minCalls": 1,
      "operationId": "report_progress",
      "requireSuccess": true
    },
    {
      "after": "report_progress",
      "before": "get_task_context",
      "id": "context-before-progress",
      "kind": "operation_order"
    },
    {
      "collection": "comments",
      "id": "one-progress-comment",
      "kind": "state_effect",
      "maximumAdded": 1,
      "minimumAdded": 1
    },
    {
      "id": "terminal-completed",
      "kind": "terminal",
      "status": "completed"
    },
    {
      "id": "transcript-complete",
      "kind": "transcript_complete"
    },
    {
      "id": "mock-only",
      "kind": "network",
      "realPaperclipRequests": 0
    }
  ],
  "description": "The agent must orient itself before recording a progress mutation on its assigned mock task.",
  "fixture": "../fixtures/companies/orientation.json",
  "id": "context-before-action",
  "prompt": "Please begin work on your assigned task. Record a brief progress update describing the immediate next step you will take.",
  "schema": "paperclip-runner/eval-case/v1",
  "title": "Agent reads context before changing task state"
}