AI Memory Hub
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MEMORY_WRITER | Yes | Identity of the AI tool or client writing to the memory (e.g., claude, codex, gemini, qwen, kimi, hermes, chatgpt, user, other). Used to tag the source of each memory. | |
| AI_MEMORY_VAULT | Yes | Path to the Obsidian vault or plain folder that stores the memory Markdown files. This is the canonical memory store. | |
| MEMORY_WRITE_MODE | No | Write mode for proposed memories. 'review' queues proposals for manual approval in the dashboard; 'auto' writes valid proposals directly to the vault. | review |
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": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| memory_policyA | Return the automatic persistent-memory retention policy and active write mode. |
| memory_searchB | Search persistent memory. Prefer this before opening files. |
| memory_readA | Read exactly one memory Markdown file. Path must be inside the vault. |
| memory_proposeC | Validate a durable memory candidate. Auto mode stores it; review mode queues it. |
| memory_supersedeB | Supersede an old memory. In review mode the replacement is queued for approval. |
| memory_forgetA | Delete one persistent memory by its stable memory ID. |
| memory_auditA | Check file/index integrity without modifying memory. |
| memory_reindexA | Rebuild the disposable SQLite search index from the Markdown vault. |
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 8 tools
Each tool targets a distinct operation: policy introspection, search, file read, write/propose, update/supersede, delete, integrity audit, and index rebuild. Memory_search vs memory_read and propose vs supersede are related but clearly separated by descriptions.
All tools share a consistent memory_ prefix and use clear snake_case action suffixes such as search, read, propose, supersede, forget, audit, and reindex. The pattern is predictable and uniform.
Eight tools is well within the ideal range and appropriate for a persistent memory hub. Each tool covers a distinct lifecycle or maintenance need without unnecessary bloat.
The core memory lifecycle is well covered: policy, search, read, create/propose, update/supersede, delete, audit, and reindex. The only minor gap is that review-mode approvals/rejections have no dedicated tool, though this is likely handled outside the agent workflow.