Free MCP tool error diagnosis, argument checks, and related agent-tool discovery.
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.
Classify a failed MCP/API tool call and recommend bounded next actions. Offline deterministic analysis. Does NOT retry, contact a service, or store the supplied error text.
True ONLY when the tool call is safe to repeat without side effects.
Error or exception text; remove secrets before sharing.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http agent-repair https://agent-repair.saromy-export.workers.dev/mcp
Classify a failed MCP/API tool call and recommend bounded next actions. Offline deterministic analysis. Does NOT retry, contact a service, or store the supplied error text.
| Argument | Type | Description |
|---|---|---|
| idempotent | boolean | True ONLY when the tool call is safe to repeat without side effects. |
| http_status | integer | |
| jsonrpc_code | integer | |
| error_message | string | Error or exception text; remove secrets before sharing. |
| retry_after_seconds | integer |
{
"type": "object",
"required": [],
"properties": {
"idempotent": {
"type": "boolean",
"description": "True ONLY when the tool call is safe to repeat without side effects."
},
"http_status": {
"type": "integer",
"maximum": 599,
"minimum": 100
},
"jsonrpc_code": {
"type": "integer"
},
"error_message": {
"type": "string",
"maxLength": 2000,
"description": "Error or exception text; remove secrets before sharing."
},
"retry_after_seconds": {
"type": "integer",
"maximum": 3600,
"minimum": 0
}
},
"additionalProperties": false
}Discover useful independent agent-native tools from this same project and their MCP connection URLs. No request to another service, referrals only when relevant.
| Argument | Type | Description |
|---|---|---|
| need | string |
{
"type": "object",
"properties": {
"need": {
"enum": [
"diagnose_failure",
"validate_arguments",
"audit_contracts",
"inspect_demo_health",
"dependency_risk",
"all"
],
"type": "string"
}
},
"additionalProperties": false
}Check supplied tool arguments against a constrained subset of JSON Schema (type, required, properties, additionalProperties, enum, const, numeric/string bounds, arrays). Returns paths and errors, no network access or storage. NOT full JSON Schema validation.
| Argument | Type | Description |
|---|---|---|
| schema* | object | MCP tool inputSchema. Limited JSON Schema subset. |
| arguments* | object | Tool call arguments. Remove credentials; no raw payload is logged. |
{
"type": "object",
"required": [
"schema",
"arguments"
],
"properties": {
"schema": {
"type": "object",
"description": "MCP tool inputSchema. Limited JSON Schema subset."
},
"arguments": {
"type": "object",
"description": "Tool call arguments. Remove credentials; no raw payload is logged."
}
},
"additionalProperties": false
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 15, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.