Memory MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| LOG_LEVEL | No | Log level: debug, info, warn, error | info |
| QDRANT_URL | No | Qdrant endpoint | http://localhost:6333 |
| EMBEDDING_URL | No | Embedding service URL (for Ollama/Custom) | |
| EMBEDDING_MODEL | No | Model name (provider default if not set) | |
| EMBEDDING_API_KEY | No | API key for OpenAI | |
| VECTORDB_PROVIDER | No | Storage provider: local, qdrant | local |
| EMBEDDING_PROVIDER | No | Embedding provider: transformersjs, openai, ollama, custom | transformersjs |
| VECTORDB_COLLECTION | No | Collection name | memories |
| EMBEDDING_DIMENSIONS | No | Vector dimensions (provider default if not set) | |
| EMBEDDING_MAX_TOKENS | No | Max token context window for embeddings (provider default if not set) |
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 |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| memory_addA | Store a memory with auto-categorization and importance scoring. Use this to persist decisions, findings, debug insights, infrastructure details, user preferences, and session summaries. |
| memory_linkA | Link two related memories bidirectionally. Memories with similarity > 0.7 are auto-linked on store, but use this for connections the auto-linker missed. |
| memory_searchA | Semantic search across all memories. Use this at the start of a session, before making decisions, or when you need context about a topic. Returns results ranked by relevance. |
| memory_listA | Browse memories by type, tags, or project with pagination. Use this when you need to review all memories in a category rather than searching by meaning. |
| memory_forgetA | Delete a memory by ID. Memories are never automatically deleted — use this for outdated or incorrect information. |
| memory_profileA | Key-value store for user preferences and settings. Use this for structured data like preferred_language, editor, timezone — not for general knowledge (use memory_add for that). |
| memory_helpA | Show server configuration and available 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 | |
TDQS
Scored across 7 tools
Each tool targets a distinct operation: add, link, search, list, forget, profile, and help. The search vs list distinction is explicitly described as semantic versus metadata browsing, and memory_profile is clearly separated from general knowledge storage.
All tools share the memory_ prefix, which is highly predictable. Most suffixes are verb-style actions like add, link, search, list, and forget, though profile and help are nouns rather than actions, creating a minor inconsistency.
Seven tools is well-scoped for a memory server: core CRUD, linking, browsing, preference storage, and help each have a clear place. There is no apparent redundancy or excessive surface area.
The core memory lifecycle is covered: add, read/search/list, delete, link, and profile storage. An explicit update operation and an unlink operation are missing, but these can be worked around with forget+add, so they are minor gaps rather than critical dead ends.