ollama-web-tools-mcp
MCP server exposing web_search, web_fetch, and cluesift_domain_fetch tools with configurable Ollama API keys and HTTP or stdio transport.
README
Ollama Web Tools MCP
Python 3.12+ MCP server preserving the Ruby oracle's web_search and web_fetch
contracts and adding cluesift_domain_fetch. It uses the official Python MCP SDK,
stdio by default, and stateless JSON streamable HTTP at / with --http.
Configuration
Production (prod or production in the first set of ENVIRONMENT,
CLUESIFT_ENV, or ENV) requires OLLAMA_KEY_POOL_FILE. The reloadable UTF-8
JSON document has exactly this useful shape; array order is strict priority and
duplicates are removed while preserving first occurrence:
{"keys":["primary-ollama-tools-key","secondary-ollama-tools-key"]}
The file is re-read when its mtime changes or after a two-second TTL. Outside
production only, comma-separated OLLAMA_CLOUD_API_KEYS, then OLLAMA_API_KEY,
are fallback inputs. Keys are never logged; diagnostics identify them only by the
first 12 lowercase hex characters of SHA-256.
Optional Ollama timeouts: OLLAMA_OPEN_TIMEOUT=10, OLLAMA_READ_TIMEOUT=60,
OLLAMA_WRITE_TIMEOUT=30. ClueSift aliases use complete endpoint URLs (not base
URLs):
cluesift-test:CLUESIFT_ARTIFACT_TEST_URLand_TOKENcluesift-prod:CLUESIFT_ARTIFACT_PROD_URLand_TOKEN
Each pair must be wholly present or absent. Callback timeout defaults are
CLUESIFT_ARTIFACT_OPEN_TIMEOUT=2, ...READ_TIMEOUT=5, and
...WRITE_TIMEOUT=5. Callback failures are content-free telemetry and never
replace a successful fetch result.
Callback URLs are complete absolute HTTP(S) endpoints, may use only their
scheme's default port, and may not contain userinfo, a query, or a fragment.
Fetched public URLs reject trailing-dot hosts rather than canonicalizing them.
Run and develop
python -m venv .venv
.venv/bin/pip install -e '.[dev]'
.venv/bin/ollama-web-tools-mcp
.venv/bin/ollama-web-tools-mcp --http --port 8767
.venv/bin/ruff check .
.venv/bin/ruff format --check .
.venv/bin/pytest
The three tools are exactly web_search, web_fetch, and
cluesift_domain_fetch. The ClueSift tool accepts only url, artifact_target,
request_id, and case_id; see the copied implementation brief for its full
security and delivery contract.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.