McpSelfTest
Runs a sequence of internal checks that confirm the transport is healthy and tool categories are reachable. Pair with GetMcpCapabilities at the start of a session.
| Property | Value |
|---|---|
| Required scope | mcp:read |
| Risk | None. Read-only. |
| Side effects | None. |
When an agent should use it
- At the start of a session, after
GetMcpCapabilities. - After a long pause when the connection state is unclear.
- Before launching an automated loop that depends on a tool group.
When an agent should not use it
- In every tool call. Once per session is enough.
Parameters
None.
Response shape
{
"transport": "ok",
"categories": {
"accounts": "ok",
"orders": "ok",
"positions": "ok",
"ninjascript": "ok",
"backtest": "ok",
"alerts": "ok"
},
"warnings": []
}
Example user prompt
Call McpSelfTest and summarize. If any category is not "ok", stop and tell me
which one and why.
Example tool call
{ "name": "McpSelfTest", "arguments": {} }
Example response
{
"transport": "ok",
"categories": {"accounts": "ok", "ninjascript": "ok", "backtest": "ok"},
"warnings": ["addon below recommended version"]
}
Common errors
| Symptom | Likely cause | Fix |
|---|---|---|
transport: degraded | Server-side issue. | Wait and retry. Check the status page. |
Category reports unavailable | Add-on does not support that capability. | Update the add-on. |
| Tool times out | NT8 is unresponsive. | Open NT8 and clear any modal dialog. |
Safe workflow placement
Always pair with GetMcpCapabilities. The capability call answers "what is on?" and McpSelfTest answers "is it working?". If either fails, the agent should stop and report.