Loads a txtai embeddings database packaged as tar.gz so agents can run semantic and graph search over a local knowledge base.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add kb --env TXTAI_EMBEDDINGS='<TXTAI_EMBEDDINGS>' -- uv venv --python=3.10 '#' or 3.11, 3.12, etc.~/.codex/config.toml
[mcp_servers.kb]
command = "uv"
args = ["venv", "--python=3.10", "#", "or", "3.11,", "3.12,", "etc."]
env = { TXTAI_EMBEDDINGS = "<TXTAI_EMBEDDINGS>" }
# codex mcp add kb --env TXTAI_EMBEDDINGS='<TXTAI_EMBEDDINGS>' -- uv venv --python=3.10 '#' or 3.11, 3.12, etc.~/.cursor/mcp.json
{
"mcpServers": {
"kb": {
"command": "uv",
"args": [
"venv",
"--python=3.10",
"#",
"or",
"3.11,",
"3.12,",
"etc."
],
"env": {
"TXTAI_EMBEDDINGS": "<TXTAI_EMBEDDINGS>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"kb": {
"type": "stdio",
"command": "uv",
"args": [
"venv",
"--python=3.10",
"#",
"or",
"3.11,",
"3.12,",
"etc."
],
"env": {
"TXTAI_EMBEDDINGS": "<TXTAI_EMBEDDINGS>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"kb": {
"command": "uv",
"args": [
"venv",
"--python=3.10",
"#",
"or",
"3.11,",
"3.12,",
"etc."
],
"env": {
"TXTAI_EMBEDDINGS": "<TXTAI_EMBEDDINGS>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"kb": {
"command": "uv",
"args": [
"venv",
"--python=3.10",
"#",
"or",
"3.11,",
"3.12,",
"etc."
],
"env": {
"TXTAI_EMBEDDINGS": "<TXTAI_EMBEDDINGS>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"kb": {
"command": "uv",
"args": [
"venv",
"--python=3.10",
"#",
"or",
"3.11,",
"3.12,",
"etc."
],
"env": {
"TXTAI_EMBEDDINGS": "<TXTAI_EMBEDDINGS>"
}
}
}
}Environment variables
TXTAI_EMBEDDINGS
Replace the placeholders with your own credentials and keep them out of version control.