Native Research MCP
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Native Research MCP research the latest EU AI Act compliance deadlines and save sources"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Native Research MCP
A stateful MCP server for DeepSeek-native web research and focused page reading.
The server uses the OpenAI Python SDK and the official MCP Python SDK. The configured provider executes native web tools. The server stores visible evidence and supports follow-up questions and recovery.
This project is independent of DeepSeek and OpenAI. It does not include credentials, a private endpoint, or a hosted service.
Install
Use Python 3.11 or later.
python -m venv .venv
. .venv/bin/activate
pip install "git+https://github.com/pipiwego/native-research-mcp.git@v0.1.0"Related MCP server: @kyaulabs/deepseek-websearch
Configure an MCP client
Set NATIVE_RESEARCH_API_KEY in the environment that starts the server.
Do not put credentials in a tracked file or command argument.
Use the executable path from your environment in the client's stdio MCP configuration:
{
"mcpServers": {
"native_research": {
"command": "/absolute/path/to/.venv/bin/native-research-mcp"
}
}
}Client configuration formats can differ. If the client does not inherit environment variables, use its credential store or environment-forwarding feature.
The default endpoint is https://api.deepseek.com/v1.
The default model is deepseek-flash.
A custom endpoint must support Responses streaming and DeepSeek-native web tools.
An OpenAI-compatible text endpoint alone is not sufficient.
Restart the server after you change its environment.
Environment variable | Function |
| Provider credential; no file fallback |
| Responses API base URL |
| Model identifier |
| Local evidence and state directory |
| Timeout for provider requests |
When no data directory is set, the server uses the account data directory from platformdirs.
MCP tools
Tool | Function |
| Create a task with a goal and scope |
| Read the current revision and request state |
| Extend a task through native web research |
| Read a public page with a specific focus |
| Read saved evidence without a provider request |
| Organize saved evidence without search or fetch tools |
Each tool describes its input and its place in the workflow. No separate skill or client-specific prompt is necessary.
For each follow-up, use the task ID and current revision.
Use a new question to extend the task.
Use refresh when cached evidence is too old for the question.
Architecture
MCP client
|
v
Official MCP SDK -> Research service -> Official OpenAI SDK
| |
v v
SQLite + JSONL Native provider toolsserver.pydefines MCP tools and validates input.config.pyreads environment settings.engine.pycontrols revisions, append-only history, evidence, and recovery.storage.pystores task and evidence records in SQLite.transport.pyprocesses Responses events and reuses the SDK connection pool.urls.pynormalizes source URLs and rejects explicit local addresses.
The server keeps a stable task prefix and appends new questions and visible evidence. It does not add topic classification templates to every request. When history exceeds its bound, it creates an explicit checkpoint. Detailed evidence remains available by ID.
The server records complete evidence lines before final synthesis. After output interruption, it can organize saved evidence without another web request. It does not depend on opaque signatures to restore a page.
Boundaries
A native request can contain more than one search or page read.
Token and time limits apply between requests, not inside provider execution.
Domain preferences keep out-of-domain sources with a label.
strictlimits accepted evidence, not provider-side browsing.Page text and evidence are untrusted input, not instructions.
Extracted quotations are model output, not independently verified page bytes.
Lexical URL checks do not provide DNS, redirect, or remote SSRF isolation.
A saved source URL does not prove a new claim.
Native page truncation can still need a focused follow-up.
Local state and journals can contain sensitive research content.
This stdio server is not a multi-user isolation service.
See SECURITY.md for data-handling boundaries.
Development
pip install -e ".[dev]"
pytest
ruff check src tests
python -m buildTests use synthetic data and a fake provider. They do not use credentials or make provider requests.
License
MIT. See LICENSE.
Available Tools
6 toolsdeepseek_read_urlBRead-only
Read a public page for a specific question. A new focus extends earlier evidence. Native tools can truncate long pages.
Inputs: gap (required), url (required), focus (required), research_id (required), +2 more.
deepseek_searchBRead-only
Extend the task with native web research. State the missing evidence. Set refresh for new verification instead of a cached result.
Inputs: gap (required), query (required), research_id (required), parent_revision (required), +1 more.
research_evidenceARead-onlyIdempotent
Read saved evidence by ID without a provider request. Evidence contains model extractions, not raw page bytes.
Inputs: research_id (required), evidence_ids.
research_recoverCRead-only
Recover from saved visible evidence. This operation uses a model request without search or fetch tools.
Inputs: gap (required), research_id (required), parent_revision (required).
research_startBRead-only
Start a research task with a goal and scope. Domain preferences do not restrict provider browsing. Budgets apply between requests.
Inputs: goal (required), scope (required), effort, domains, +5 more.
research_statusARead-onlyIdempotent
Read the current revision, request status, and evidence count without a provider request.
Inputs: research_id (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
The tools cover distinct research lifecycle stages: start, search, URL read, evidence read, status, and recovery. There is some potential overlap between research_evidence and research_recover, and between deepseek_read_url and research_evidence, but the descriptions clarify saved extraction versus provider fetch versus recovery.
All names use snake_case and coherent domain prefixes: research_* for task lifecycle and deepseek_* for provider actions. The only inconsistency is that some research_* names are verbs (start, recover) while others are resource nouns (evidence, status).
Six tools is well within the 3-15 range and each covers a distinct step in the research workflow. No tool appears redundant or filler.
Core research lifecycle is covered: start, search, read, inspect evidence/status, and recover. Minor gaps remain, such as no explicit finish/synthesize/cancel or list-evidence operation, but agents can work around these via status and evidence IDs.
Maintenance
Related MCP Connectors
Web search, fetch, extract, and research for AI agents. Markdown output + AI-synthesized answers.
Real-time web and scholarly search with cited answers and multi-step deep research.
Web search and page reading for AI agents, including JSON APIs, RSS feeds and YouTube.
Search the web and read pages from your AI assistant, with results it can cite.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables real-time web search via DeepSeek's search-enhanced dialogue, providing search results as answers through MCP tools.1MIT
- AlicenseAqualityDmaintenanceEnables real-time web search via DeepSeek's server-side tool without requiring additional search API keys.112 npmMIT
- FlicenseNot gradedqualityBmaintenanceEnables AI agents to perform deep web research by searching, fetching, chunking, retrieving, and synthesizing cited answers from exact source passages via a deep_research tool.-
- AlicenseAqualityBmaintenanceEnables finding information online, reading web pages, finding images, capturing screenshots, and performing multi-source research with per-answer confidence.5AGPL 3.0