Monitoring

mnemory exposes a /metrics endpoint in Prometheus text exposition format, enabled by default (ENABLE_METRICS=true).

Endpoint Access

By default, /metrics and /health are served on the main port with standard API key authentication. When MGMT_PORT is set, the same endpoints are also exposed on the management port without auth. For production, scrape and probe the management port to avoid sending credentials and to prevent duplicate ingestion (see Management Port).

Available Metrics

Counters (in-memory, reset on restart)

Metric	Labels	Description
mnemory_operations_total	operation, user_id, agent_id	Total MCP/REST operations. Operations: add_memory, add_memories, search_memories, find_memories, update_memory, delete_memory, delete_all, get_core_memories, get_recent_memories, list_memories, save_artifact, get_artifact, list_artifacts, delete_artifact, list_categories, recall, remember, initialize_memory, fsck_check, fsck_apply
mnemory_remember_facts_extracted_total	user_id	Facts extracted by remember pipeline (pre-dedup)
mnemory_remember_facts_stored_total	user_id	Facts stored by remember pipeline (post-dedup)
mnemory_remember_dedup_decisions_total	user_id, action	Remember dedup decisions (add, update, delete, skip)
mnemory_remember_empty_extractions_total	user_id	Remember calls that produced zero facts
mnemory_consolidation_runs_total	user_id, type	Consolidation runs (session, cross_session)
mnemory_consolidation_memories_produced_total	user_id	Consolidated memories produced
mnemory_consolidation_memories_superseded_total	user_id	Raw memories superseded by consolidation
mnemory_consolidation_validation_failures_total	user_id	Consolidation output validation failures

Histograms (in-memory, reset on restart)

Metric	Labels	Description
mnemory_consolidation_duration_seconds	user_id, type	Consolidation run duration (buckets: 1, 5, 10, 30, 60, 120, 300s)

Gauges (from Qdrant, cached)

Refreshed asynchronously after the configurable cache TTL expires (METRICS_CACHE_TTL, default 600s). Stale values remain available during refresh.

Metric	Labels	Description
mnemory_memories_total	user_id, agent_id, memory_type, role	Total memories by all key dimensions
mnemory_memories_decayed_total	user_id, agent_id	Decayed (expired) memories
mnemory_memories_pinned_total	user_id, agent_id	Pinned memories
mnemory_memories_by_category_total	user_id, category	Memories per category
mnemory_memories_with_artifacts_total	user_id, agent_id	Memories with artifacts attached
mnemory_active_sessions	--	Active memory sessions (recall/remember)
mnemory_info	version, vector_backend, artifact_backend	Server metadata (always 1)
mnemory_consolidation_sessions_pending	user_id	Sessions awaiting consolidation

Prometheus Scrape Config

scrape_configs:
  - job_name: mnemory
    scrape_interval: 60s
    static_configs:
      - targets: ['mnemory:9090']  # MGMT_PORT

Grafana Dashboard

A pre-built Grafana dashboard is available in integrations/grafana/. Import dashboard.json into Grafana for an overview of memories, operations, and breakdowns by type/category/role with user and agent filtering.