voice-memory MCP server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| VOICE_MEMORY_DB | No | Path to the SQLite database file. If unset, data/voice_memory.db resolved relative to the file location is used. |
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
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_voice_memoriesA | Keyword search (substring match) over title, summary, transcript and tags. Returns id, recorded_at, title, summary, transcript excerpt and tags, newest first. limit is capped at 50. |
| get_voice_memoryA | Get the full record (including transcript) for one memory by its UUID. |
| get_recent_memoriesA | List the most recent memories, newest first. limit is capped at 50. |
| list_voice_tagsA | List all tags with the number of memories using each. |
| save_voice_memoryA | Save a new memory. title (<=200 chars) and transcript (<=100000 chars) are required; summary <=2000 chars; up to 20 tags of <=50 chars each. |
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 access pattern: single-by-ID retrieval (get_voice_memory), keyword search (search_voice_memories), recency listing (get_recent_memories), tag enumeration (list_voice_tags), and write (save_voice_memory). An agent can easily choose between search vs. recent vs. get-by-UUID based on descriptions.
All tools follow a verb_noun pattern (get_, search_, list_, save_), which is consistent. Minor deviation: the noun form shifts between 'voice_memory', 'voice_memories', 'recent_memories', and 'voice_tags', which is readable but not perfectly uniform.
Five tools is well-scoped for a voice-memory store: one write, three read/query paths, and one tag-listing helper. Each tool earns its place without redundancy.
The read/create and search surface is solid, but there is no update or delete operation for memories, and no way to manage tags directly. For a memory store these are notable gaps that could cause dead ends if an agent needs to correct or remove a record.