x-search
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| X_SEARCH_MODEL | No | Grok model served by magpie. | grok-plugin/grok-4.7 |
| X_SEARCH_PYTHON | No | Python executable path for the npm launcher. Set this if Python detection fails. The value must not contain command arguments. | Automatic detection |
| X_SEARCH_WORKERS | No | Concurrent searches per process, 1 to 32. | 4 |
| X_SEARCH_CACHE_PATH | No | Shared SQLite database path. All terminals that should share searches must use the same database path. The database must be on a local filesystem. | OS cache directory plus magpie-x-search/cache.sqlite3 |
| X_SEARCH_MAGPIE_URL | No | Local magpie Responses endpoint. | http://127.0.0.1:3425/v1/responses |
| X_SEARCH_TIMEOUT_SEC | No | Total request timeout, 1 to 3600 seconds. | 150 |
| X_SEARCH_CACHE_TTL_SEC | No | Cache lifetime; 0 disables caching and shared execution. | 300 |
| X_SEARCH_MAX_TOOL_CALLS | No | Native search call limit, 1 to 100. | 5 |
| X_SEARCH_MAX_OUTPUT_TOKENS | No | Optional output token limit, 256 to 32768; omitted from requests by default. |
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 | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| x_searchA | Search public X/Twitter profiles, posts and threads using native xAI X search through local magpie OAuth. Returns a summary, structured posts, native source URLs and execution counts. Results are retrieved samples, not a complete archive. Use cache_mode=refresh for a fresh lookup. |
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 1 tool
The server exposes a single tool, so there is no possibility of misselection or overlapping purpose. Any agent request for X/Twitter profile, post, or thread search maps unambiguously to x_search.
x_search is a clean snake_case verb_noun name and mirrors the server name x-search. With only one tool there is no broader pattern to evaluate, so consistency is trivially satisfied rather than demonstrated.
One tool is at the thin end of the acceptable range; the rubric treats 1-2 tools as borderline. The scope (X/Twitter search) is narrow enough that a single tool is defensible, but it provides no alternative operations.
Search is covered with structured output, URLs, and cache control, but notable gaps exist: no fetch-by-post-ID, no profile lookup, no thread expansion, and no pagination or result-limit controls beyond cache_mode. Agents can work around this only by re-querying search.