Checkpoint-paired twin trials for agent fleet changes with blinded verdicts.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Fork a checkpoint-paired assignment for one task (blinded arms A/B).
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Fork a checkpoint-paired assignment for one task (blinded arms A/B).
| Argument | Type | Description |
|---|---|---|
| task* | object | |
| trial_id* | string |
{
"type": "object",
"required": [
"trial_id",
"task"
],
"properties": {
"task": {
"type": "object"
},
"trial_id": {
"type": "string"
}
}
}Register a hash-bound trial protocol; twins fork from the identical checkpoint, exactly one registered variable may differ.
| Argument | Type | Description |
|---|---|---|
| protocol* | object | |
| checkpoint* | object | |
| treated_config* | object |
{
"type": "object",
"required": [
"protocol",
"checkpoint",
"treated_config"
],
"properties": {
"protocol": {
"type": "object"
},
"checkpoint": {
"type": "object"
},
"treated_config": {
"type": "object"
}
}
}Run one pair with the deterministic scripted agent and score blind. $0.25/pair, first 10 free once per operator.
| Argument | Type | Description |
|---|---|---|
| task* | object | |
| trial_id* | string |
{
"type": "object",
"required": [
"trial_id",
"task"
],
"properties": {
"task": {
"type": "object"
},
"trial_id": {
"type": "string"
}
}
}Alias of trial_submit_outputs (blinded scoring).
| Argument | Type | Description |
|---|---|---|
| task* | object | |
| trial_id* | string | |
| control_output* | any | |
| treated_output* | any |
{
"type": "object",
"required": [
"trial_id",
"task",
"control_output",
"treated_output"
],
"properties": {
"task": {
"type": "object"
},
"trial_id": {
"type": "string"
},
"control_output": {},
"treated_output": {}
}
}Submit the operator twins' outputs for one pair; blinded deterministic scoring. $0.25/pair, first 10 free once per operator.
| Argument | Type | Description |
|---|---|---|
| task* | object | |
| trial_id* | string | |
| control_output* | any | |
| treated_output* | any |
{
"type": "object",
"required": [
"trial_id",
"task",
"control_output",
"treated_output"
],
"properties": {
"task": {
"type": "object"
},
"trial_id": {
"type": "string"
},
"control_output": {},
"treated_output": {}
}
}Unblind after the verdict; hashed signed receipt covering every pair.
| Argument | Type | Description |
|---|---|---|
| trial_id* | string |
{
"type": "object",
"required": [
"trial_id"
],
"properties": {
"trial_id": {
"type": "string"
}
}
}Trial verdict under the enforced stopping rule.
| Argument | Type | Description |
|---|---|---|
| trial_id* | string |
{
"type": "object",
"required": [
"trial_id"
],
"properties": {
"trial_id": {
"type": "string"
}
}
}