universal-memory-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MEMORY_DATABASE_PATH | No | SQLite database path | ./memory.db |
| MEMORY_KEYWORD_WEIGHT | No | Hybrid search keyword weight | 0.4 |
| MEMORY_EMBEDDING_MODEL | No | HuggingFace model name | sentence-transformers/all-MiniLM-L6-v2 |
| MEMORY_SEMANTIC_WEIGHT | No | Hybrid search semantic weight | 0.6 |
| MEMORY_LLAMA_SERVER_URL | No | llama-server endpoint | http://localhost:8787 |
| MEMORY_EMBEDDING_BACKEND | No | transformers or llama-server | transformers |
| MEMORY_ENABLE_EMBEDDINGS | No | Set false for keyword-only search | true |
| MEMORY_EMBEDDING_DIMENSION | No | Embedding vector size | 384 |
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
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tasks | {
"list": {},
"cancel": {},
"requests": {
"tools": {
"call": {}
},
"prompts": {
"get": {}
},
"resources": {
"read": {}
}
}
} |
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| recall_memoriesA | Retrieve memories: hybrid/keyword/semantic search (query), exact fetch (memory_id), or entity lookup (entity). Exactly one selector — query, memory_id, or entity — must be provided. Set expand_links>0 to include graph neighbors for each result. |
| store_memoryB | Store a new memory with optional agent/session scoping, semantic embedding, and graph links. |
| update_memoryB | Update a memory with optimistic locking for concurrent agent safety. |
| manage_sessionA | Manage sessions for checkpoint/restore of agent state across conversations. |
| memory_adminC | Administrative operations: statistics, background jobs, entity/contradiction maintenance, manual graph links, and deletion. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 5 tools
Each tool has a clearly distinct purpose: recall, store, update, session management, and admin operations. There is no overlap between their scopes, and the descriptions make the boundaries explicit.
Tool names follow a consistent verb_noun pattern (e.g., store_memory, update_memory, manage_session). The only minor deviation is memory_admin which uses a noun_verb pattern, but it remains clear and readable.
With 5 tools, the set is well-scoped for a memory server. Each tool covers an essential operation—CRUD for memories, session management, and admin—without being excessively minimal or bloated.
The tool set provides complete lifecycle coverage for memories (store, retrieve, update, delete via admin) plus session management and administrative maintenance. No obvious gaps are present for the stated purpose of agent memory management.