Skip to main content
Glama
Zhu-Guang-Scion

duckseek

by Zhu-Guang-Scion
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
EMB_MODELYesModel name for the embedding model.
LLM_MODELYesModel name for the LLM.
EMB_API_KEYYesAPI key for the embedding provider.
LLM_API_KEYYesAPI key for the LLM provider.
EMB_BASE_URLYesBase URL for the embedding API (OpenAI compatible).
LLM_BASE_URLYesBase URL for the LLM API (OpenAI compatible).
NL2DATA_CONFIGNoOptional path to override the default config.yaml location.

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

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
duckseek_statusA

Environment + data readiness for DuckSeek.

Reports which of the six env vars (LLM_BASE_URL/LLM_API_KEY/LLM_MODEL, EMB_BASE_URL/EMB_API_KEY/EMB_MODEL) are missing — names only, never values — plus registered sources/tables and retrieval-index state. Call this first; if variables are missing, ask the user to provide them into the server environment.

duckseek_list_tablesA

List registered tables grouped by source, with row counts.

Row counts come from stored profiles; rows: null means the table has not been profiled yet. Read-only.

duckseek_askA

Answer a natural-language question over the registered tables.

Runs retrieve → SQL generation → read-only guard → sandboxed execution. Returns answer (compact markdown of the result), sql (the executed statement, for verification), row_count, elapsed_ms, source_tables and embedding_degraded (true when this call's retrieval fell back to BM25-only — tell the user and suggest providing EMB_* for best recall); interpret the answer yourself. Ambiguous questions return needs_clarification with a follow-up question to ask the user. Read-only; the SQL is verified to be a single SELECT before it runs.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.4/5.0

Scored across 3 tools

Disambiguation4/5

duckseek_status and duckseek_list_tables both surface registered tables and index state, creating mild overlap, but status is clearly a readiness check while list_tables is metadata-only. duckseek_ask is fully distinct as the query entry point.

Naming Consistency4/5

All tools share the duckseek_ prefix and snake_case convention, but the structure is mixed: status is a noun, list_tables is verb_noun, and ask is verb-only. The pattern is still predictable and readable.

Tool Count5/5

Three tools cover the complete workflow: check readiness, inspect available tables, and ask a natural-language question. Each tool earns its place and the small count fits the server's narrow read-only querying purpose.

Completeness4/5

The core lifecycle of status discovery, table inspection, querying, and clarification is present, with good guardrails like read-only SQL verification and BM25 fallback reporting. The main gap is that list_tables exposes row counts but not schemas or columns, so agents must infer schema details through duckseek_ask.

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources