Gemdex MCP
Gemdex is a persistent memory layer for AI coding agents that allows saving, recalling, and updating memories across repos and sessions. It uses Gemma embeddings and embedded LanceDB for hybrid semantic and BM25 retrieval, enabling agents to index small chunks and resolve matches to whole parent memories for precise context retrieval.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
What you can do with it
- Save durable memories for later recall across sessions
- Recall relevant memories using text queries or media examples
- Update existing memories to correct or refine stored information
- Delete memories that are no longer needed
Source and maintenance
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
gemdex-mcp 1.0.47
{
"version": "1.0.47",
"transport": "stdio",
"identifier": "gemdex-mcp",
"runtimeHint": "npx",
"registryType": "npm",
"runtimeArguments": [
{
"type": "positional",
"value": "-y",
"valueHint": "auto-confirm install"
},
{
"type": "positional",
"value": "gemdex-mcp@1.0.47",
"valueHint": "package spec"
}
],
"environmentVariables": [
{
"name": "GEMINI_API_KEY",
"format": "string",
"isRequired": true,
"description": "Google AI Studio key for local mode, or the exact value 'local' for on-device MLX text embeddings. Not needed in remote mode."
},
{
"name": "GEMDEX_MODE",
"format": "string",
"choices": [
"local",
"remote"
],
"default": "local",
"description": "Memory backend: embedded local LanceDB or a self-hosted Gemdex Server"
},
{
"name": "LANCEDB_PATH",
"format": "filepath",
"default": "~/.gemdex/lance",
"description": "Directory for the embedded LanceDB store"
},
{
"name": "EMBEDDING_MODEL",
"format": "string",
"choices": [
"gemini-embedding-2",
"gemini-embedding-001"
],
"default": "gemini-embedding-2",
"description": "Gemini text embedding model"
},
{
"name": "GEMDEX_REMOTE_URL",
"format": "string",
"description": "Gemdex Server root URL (required in remote mode)"
},
{
"name": "GEMDEX_REMOTE_TOKEN",
"format": "string",
"description": "Gemdex Server bearer token (required in remote mode)"
},
{
"name": "GEMDEX_REMOTE_NAME",
"format": "string",
"description": "Human-readable name for the remote server"
},
{
"name": "GEMDEX_TRUST_RANKING",
"format": "boolean",
"default": "false",
"description": "Re-rank recall results by recorded worked/failed/stale outcomes"
}
]
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"gemdex-mcp": {
"command": "npx",
"args": [
"-y",
"gemdex-mcp@1.0.47"
],
"env": {
"GEMINI_API_KEY": "<YOUR_GEMINI_API_KEY>",
"GEMDEX_MODE": "<YOUR_GEMDEX_MODE>",
"LANCEDB_PATH": "<YOUR_LANCEDB_PATH>",
"EMBEDDING_MODEL": "<YOUR_EMBEDDING_MODEL>",
"GEMDEX_REMOTE_URL": "<YOUR_GEMDEX_REMOTE_URL>",
"GEMDEX_REMOTE_TOKEN": "<YOUR_GEMDEX_REMOTE_TOKEN>",
"GEMDEX_REMOTE_NAME": "<YOUR_GEMDEX_REMOTE_NAME>",
"GEMDEX_TRUST_RANKING": "<YOUR_GEMDEX_TRUST_RANKING>"
}
}
}
}Cline
{
"mcpServers": {
"gemdex-mcp": {
"command": "npx",
"args": [
"-y",
"gemdex-mcp@1.0.47"
],
"env": {
"GEMINI_API_KEY": "<YOUR_GEMINI_API_KEY>",
"GEMDEX_MODE": "<YOUR_GEMDEX_MODE>",
"LANCEDB_PATH": "<YOUR_LANCEDB_PATH>",
"EMBEDDING_MODEL": "<YOUR_EMBEDDING_MODEL>",
"GEMDEX_REMOTE_URL": "<YOUR_GEMDEX_REMOTE_URL>",
"GEMDEX_REMOTE_TOKEN": "<YOUR_GEMDEX_REMOTE_TOKEN>",
"GEMDEX_REMOTE_NAME": "<YOUR_GEMDEX_REMOTE_NAME>",
"GEMDEX_TRUST_RANKING": "<YOUR_GEMDEX_TRUST_RANKING>"
}
}
}
}VS Code
{
"servers": {
"gemdex-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"gemdex-mcp@1.0.47"
],
"env": {
"GEMINI_API_KEY": "<YOUR_GEMINI_API_KEY>",
"GEMDEX_MODE": "<YOUR_GEMDEX_MODE>",
"LANCEDB_PATH": "<YOUR_LANCEDB_PATH>",
"EMBEDDING_MODEL": "<YOUR_EMBEDDING_MODEL>",
"GEMDEX_REMOTE_URL": "<YOUR_GEMDEX_REMOTE_URL>",
"GEMDEX_REMOTE_TOKEN": "<YOUR_GEMDEX_REMOTE_TOKEN>",
"GEMDEX_REMOTE_NAME": "<YOUR_GEMDEX_REMOTE_NAME>",
"GEMDEX_TRUST_RANKING": "<YOUR_GEMDEX_TRUST_RANKING>"
}
}
}
}Claude Desktop
{
"mcpServers": {
"gemdex-mcp": {
"command": "npx",
"args": [
"-y",
"gemdex-mcp@1.0.47"
],
"env": {
"GEMINI_API_KEY": "<YOUR_GEMINI_API_KEY>",
"GEMDEX_MODE": "<YOUR_GEMDEX_MODE>",
"LANCEDB_PATH": "<YOUR_LANCEDB_PATH>",
"EMBEDDING_MODEL": "<YOUR_EMBEDDING_MODEL>",
"GEMDEX_REMOTE_URL": "<YOUR_GEMDEX_REMOTE_URL>",
"GEMDEX_REMOTE_TOKEN": "<YOUR_GEMDEX_REMOTE_TOKEN>",
"GEMDEX_REMOTE_NAME": "<YOUR_GEMDEX_REMOTE_NAME>",
"GEMDEX_TRUST_RANKING": "<YOUR_GEMDEX_TRUST_RANKING>"
}
}
}
}Let your agent find its tools
Find servers like Gemdex MCP from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.