Skip to content
Docs menu

Tool reference · Maintenance & performance

memory_perf_report

Returns the server’s internal timing and call counters and embedding-cache statistics.

local stdio server read-only

When to use

  • Searches got slower and you want to see where the time goes.
  • You want to confirm the fast path made no LLM or network calls.
  • You want to see how well the embedding cache is being used.

Parameters

This tool takes no parameters.

Example

Arguments

{}

Result shape

{
  "counters": {
    "llm_calls": 0,
    "network_calls": 0,
    "fts_ms": 212.4,
    "embed_ms": 988.1,
    "vector_ms": 140.7,
    "save_total_ms": 1530.2,
    "search_total_ms": 1402.9
  },
  "embedding_cache_v11": {
    "rows": 5120,
    "total_hits": 18344,
    "by_space": {
      "text": 4870,
      "code": 250
    }
  }
}

Counters are totals since the process started. If a part fails, a counters_error or embedding_cache_v11_error key appears instead.

Values are illustrative; the keys follow the server's handler. MCP clients receive the result as JSON text content.

Server description

The description the server sends to your agent in tools/list, captured from the v14.7.0 source:

dump in-process telemetry counters (search_total_ms, embed_ms, fts_ms, vector_ms, llm_calls, network_calls) plus persistent embedding_cache stats. Use to verify the fast hot path stays clean.

Found a mistake? Open an issue on GitHub.

Search