Record phase two acceptance evidence and synchronize delivery status

This commit is contained in:
2026-09-07 02:55:19 +08:00
parent 89df10bc4e
commit e667dd55dd
24 changed files with 13892 additions and 45 deletions
@@ -0,0 +1,196 @@
{
"provider_id": "provider_65fd326edba646e3bc735c25fbf5ffe8",
"protocol": "openai_compatible",
"model": "deepseek-v4-flash",
"checks": {
"discovery": {
"passed": true,
"count": 3,
"elapsed_ms": 1904.77
},
"normal_chat": {
"passed": true,
"input_tokens": 87,
"output_tokens": 16,
"elapsed_ms": 2107.22
},
"stream_usage_reasoning": {
"passed": true,
"events": [
"Done",
"TextDelta",
"ThinkingDelta",
"Usage"
],
"usage_reported": true,
"reasoning_observed": true,
"elapsed_ms": 2118.0
},
"stream_cancel": {
"passed": true,
"first_event": "ThinkingDelta",
"close_ms": 12.160799989942461,
"scope": "client stream resource close; provider billing cessation not observable",
"elapsed_ms": 1842.69
},
"error_mapping": {
"passed": true,
"error_code": "PROVIDER_INVALID_REQUEST",
"elapsed_ms": 1590.6
}
},
"unconfigured_protocols": [
"openai_responses",
"anthropic_messages",
"ollama"
],
"not_tested": [
"vendor_maximum_context_boundary"
],
"max_direct_requests": 5,
"agent": {
"run_id": "benchmark_8ebbf8c6f27f",
"kind": "agent",
"dataset_id": "agent-core-v1",
"dataset_hash": "sha256:82fd1b84cc2dc2feec0fc67357d64b145dcdf6c44ec2660f182d550e85b9d27f",
"status": "completed",
"config_snapshot": {
"dataset_id": "agent-core-v1",
"provider_id": "provider_65fd326edba646e3bc735c25fbf5ffe8",
"model": "deepseek-v4-flash",
"max_steps": 6,
"timeout_seconds": 90,
"token_budget": 6000,
"repeat": 1,
"allow_network": false,
"offline": false,
"dataset_hash": "sha256:82fd1b84cc2dc2feec0fc67357d64b145dcdf6c44ec2660f182d550e85b9d27f",
"dataset_version": "1.0.0",
"execution": "real_agent_runtime",
"provider_type": "openai_compatible",
"scoring_version": "1.0",
"permission_policy": "runtime_user_decision",
"active_agent_run_id": "run_c9296aaea4c7429db97528edc72b6180"
},
"metrics": {
"total_cases": 4,
"task_success_rate": 1.0,
"tool_selection_accuracy": 1.0,
"tool_argument_accuracy": 1.0,
"invalid_tool_call_rate": 0.0,
"average_steps": 1.75,
"average_latency_ms": 5682.329325005412,
"token_usage": 5888,
"tool_calls": 3,
"expected_calls": 3
},
"cases": [
{
"case_id": "arithmetic",
"repeat": 0,
"agent_run_id": "run_480c159ffc0341f9883a09914ff003b5",
"success": true,
"tool_calls": 1,
"expected_calls": 1,
"selected_calls": 1,
"accurate_calls": 1,
"invalid_calls": 0,
"steps": 2,
"latency_ms": 4888.349500019103,
"token_usage": 1545,
"checks": {
"completed": true,
"tools_selected": true,
"tool_arguments": true,
"no_extra_calls": true,
"tool_results": true,
"output": true,
"citation": true,
"tasks_created": true
},
"error_code": null
},
{
"case_id": "echo",
"repeat": 0,
"agent_run_id": "run_4fdbd0508bb84bcda424d97de1b91241",
"success": true,
"tool_calls": 1,
"expected_calls": 1,
"selected_calls": 1,
"accurate_calls": 1,
"invalid_calls": 0,
"steps": 2,
"latency_ms": 4978.496600000653,
"token_usage": 1601,
"checks": {
"completed": true,
"tools_selected": true,
"tool_arguments": true,
"no_extra_calls": true,
"tool_results": true,
"output": true,
"citation": true,
"tasks_created": true
},
"error_code": null
},
{
"case_id": "no-tool",
"repeat": 0,
"agent_run_id": "run_5dfb4ae0a6c547b79706ad102e577bab",
"success": true,
"tool_calls": 0,
"expected_calls": 0,
"selected_calls": 0,
"accurate_calls": 0,
"invalid_calls": 0,
"steps": 1,
"latency_ms": 2549.696500005666,
"token_usage": 541,
"checks": {
"completed": true,
"tools_selected": true,
"tool_arguments": true,
"no_extra_calls": true,
"tool_results": true,
"output": true,
"citation": true,
"tasks_created": true
},
"error_code": null
},
{
"case_id": "markdown-catalog",
"repeat": 0,
"agent_run_id": "run_c9296aaea4c7429db97528edc72b6180",
"success": true,
"tool_calls": 1,
"expected_calls": 1,
"selected_calls": 1,
"accurate_calls": 1,
"invalid_calls": 0,
"steps": 2,
"latency_ms": 10312.774699996226,
"token_usage": 2201,
"checks": {
"completed": true,
"tools_selected": true,
"tool_arguments": true,
"no_extra_calls": true,
"tool_results": true,
"output": true,
"citation": true,
"tasks_created": true
},
"error_code": null
}
],
"error": null,
"error_code": null
},
"followup_evidence": [
"context.json",
"integration.json"
]
}