turing_AgentMemory_MCP
by Chetto983
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| EMBED_MODEL | No | Embedding model name. | |
| ARCADEDB_URL | Yes | URL of the ArcadeDB server that the MCP service connects to. | |
| BERTONI_HOME | No | Path for MCP app-state such as document job queue, staging, tenant registry, and audit/span JSONL. Compose mounts this under /bertoni. | |
| GLINER_MODEL | No | GLiNER model identifier. | |
| RERANK_BLEND | No | Set to 1 to blend fused retrieval and provider orders with reciprocal-rank fusion; set to 0 for guarded pure rerank ordering. Defaults to 1. | |
| RERANK_MODEL | No | Rerank model name. | |
| EMBED_API_KEY | No | API key for the embedding provider. Overrides PROVIDER_API_KEY. | |
| GLINER_LABELS | No | Entity labels for GLiNER extraction. | |
| GLINER_REDACT | No | Set to 1 to redact detected entity spans before storage and embedding. | |
| EMBED_BASE_URL | No | OpenAI-compatible embeddings base URL. Local non-Docker default is http://127.0.0.1:8081. | |
| GLINER_BACKEND | No | Entity extraction backend, for example gliner, gliner2, gliner2_onnx, or gliner2_http. | |
| GLINER_ENABLED | No | Enable optional GLiNER/GLiNER2 entity detection. | |
| RERANK_API_KEY | No | API key for the rerank provider. Overrides PROVIDER_API_KEY. | |
| GLINER_BASE_URL | No | Base URL for the GLiNER HTTP backend. | |
| RERANK_BASE_URL | No | OpenAI-compatible rerank base URL. Local non-Docker default is http://127.0.0.1:8085. | |
| EMBED_DIMENSIONS | No | Embedding vector dimensions. | |
| GLINER_PRECISION | No | GLiNER precision setting. | |
| GLINER_PROVIDERS | No | GLiNER providers setting. | |
| GLINER_THRESHOLD | No | GLiNER entity detection threshold. | |
| PROVIDER_API_KEY | No | Shared fallback API key when embedding and rerank use the same cloud provider key. | |
| RERANK_THRESHOLD | No | Rerank threshold setting. | |
| RERANK_DIMENSIONS | No | Rerank provider dimensions setting. | |
| EMBED_TIMEOUT_SECONDS | No | Timeout in seconds for embedding provider calls. | |
| AGENTMEMORY_AUTH_TOKEN | No | Static MCP bearer token for HTTP/SSE clients. Leave unset for local stdio or unauthenticated development. | |
| GLINER_TIMEOUT_SECONDS | No | Timeout in seconds for GLiNER provider calls. | |
| RERANK_TIMEOUT_SECONDS | No | Timeout in seconds for rerank provider calls. | |
| AGENTMEMORY_AUDIT_JSONL | No | Path for structured audit JSONL events without content/text/query payloads. | |
| AGENTMEMORY_AUTH_SCOPES | No | FastMCP static-token scopes metadata. | |
| AGENTMEMORY_AUTH_TOKENS | No | Comma-separated bearer tokens for MCP token rotation. Alternative to AGENTMEMORY_AUTH_TOKEN. | |
| PROVIDER_API_KEY_HEADER | No | Header used for provider API key auth. Defaults to Authorization. | |
| PROVIDER_API_KEY_SCHEME | No | Scheme used for provider API key auth. Defaults to Bearer; set empty for raw key headers. | |
| AGENTMEMORY_UTCP_AUTH_ENV | No | Environment variable name referenced by the generated UTCP manual for auth. | |
| RERANK_PROVIDER_MIN_SCORE | No | Minimum trusted rerank provider top score. Defaults to 0. | |
| AGENTMEMORY_AUTH_CLIENT_ID | No | FastMCP static-token client ID metadata. | |
| AGENTMEMORY_FUSION_WEIGHTS | No | JSON object overriding weighted reciprocal-rank fusion weights. | |
| RERANK_PRESERVE_SEED_MARGIN | No | Margin for preserving the top hybrid seed when RERANK_BLEND=0. Defaults to 0.05. | |
| AGENTMEMORY_UTCP_MCP_COMMAND | No | JSON command used in the generated UTCP manual for MCP-backed tools. | |
| AGENTMEMORY_UTCP_SERVER_NAME | No | Optional server name for the generated UTCP manual. | |
| AGENTMEMORY_DOCUMENT_JOB_PATH | No | Path for the durable document job SQLite database. | |
| AGENTMEMORY_REDACTION_ENABLED | No | Set to 1 to enable built-in secret/API-key/email pattern redaction before graph writes and vector embedding. | |
| AGENTMEMORY_OBSERVABILITY_JSONL | No | Path for timing span JSONL events for embed, ArcadeDB query, vector load, rerank, chunking, and MCP tool latency. | |
| AGENTMEMORY_AUTH_REQUIRED_SCOPES | No | Required scopes for MCP auth checks. | |
| AGENTMEMORY_DOCUMENT_STAGING_ROOT | No | Root directory for durably staged document ingest files. | |
| AGENTMEMORY_DOCUMENT_JOB_MAX_ATTEMPTS | No | Maximum attempts for a document ingest job. | |
| AGENTMEMORY_DOCUMENT_JOB_POLL_SECONDS | No | Document job polling interval in seconds. | |
| AGENTMEMORY_DOCUMENT_JOB_LEASE_SECONDS | No | Document job lease duration in seconds. | |
| AGENTMEMORY_DOCUMENT_JOB_HEARTBEAT_SECONDS | No | Document job heartbeat interval in seconds. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Server capabilities have not been inspected yet.
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
No tools | |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
This server cannot be deployed
Maintenance
ActivityInactive
ResponsivenessNo issues