shared-agent-memory-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| NOTION_TOKEN | Yes | Your Notion integration token (secret_your_token_here). | |
| NOTION_DATABASE_ID | Yes | The ID of the Notion database shared with your integration. | |
| OBSIDIAN_VAULT_PATH | No | Optional path to a Git-backed Obsidian Vault for dual-write sync. If not set or folder does not exist, the server runs in Notion-only mode. |
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_searchA | Search the shared long-term memory stored in Notion (used by Cline, OpenCode, Claude Code, GitHub Copilot, Hermes). Returns matching memories with id, title, content, agent, category, tags. Semantic and hybrid vector matches also include match.excerpt with Unicode start/end offsets. Set response to compact to omit full content and return an excerpt; use memory_get for full content. USE THIS at the start of a task with relevant keywords, and before making decisions, to reuse saved preferences, conventions, and past decisions. |
| memory_recentA | List the most recently updated memories from the shared Notion memory. Set response to compact to return excerpts without full content. Use to get context about what was learned/saved lately. |
| memory_getA | Fetch a single memory by its id (full content). |
| memory_addA | Save a new durable memory into the shared Notion memory. USE THIS when: the user states a preference, a project decision is made, a convention is established, a non-obvious bug is fixed (root cause + fix), or important project/environment context is learned. Do NOT store secrets, tokens, or throwaway information. |
| memory_updateA | Update an existing memory by id. Prefer updating over creating duplicates when knowledge changes or gets corrected. |
| memory_deleteA | Archive a memory (default) or move it to Notion trash (hard=true). Use when a memory is wrong or obsolete. |
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 6 tools
Each tool targets a distinct operation: search, recent, add, get, update, delete. There is no overlap in purpose, and the descriptions clarify when to use each, minimizing selection ambiguity.
All tools follow a consistent 'memory_' prefix followed by an action verb (search, recent, delete, add, update, get). Even 'recent' is a clear descriptor, and the pattern is uniform and predictable.
With 6 tools, the server covers the full lifecycle of memory operations—create, read, update, delete, plus search and recent listing—without being bloated or sparse. The count is well-scoped for its purpose.
The surface provides complete CRUD coverage (add, get, update, delete) and includes discovery mechanisms (search, recent). No obvious gaps exist; it fully supports the stated purpose of a shared memory store.