Research MCP Service
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Research MCP Service
Research service for search, page reading, document conversion, cached evidence, RSS and interactive Chromium/Camoufox browsing. Both local test and VPS deployments expose 40 MCP tools. The local Scrapling trial uses Scrapling as a library inside the existing worker for Chromium acquisition and adds targeted extraction, network-response inspection, action chains and element screenshots. Streamable HTTP with bearer authentication over loopback/Tailscale or HTTPS. One owner per independent deployment; clients sharing a token share its cache and browser pool. No OAuth or shared-application multi-user tenancy.
LIVE-STATUS.md: current deployment, commands, verified behavior and known limitations.
INSTALL.md: image-only Compose installation, verified image digests, startup and maintenance.
HTTPS.md: optional public TLS and separate endpoint paths.
DESIGN.md: runtime boundaries and remaining security work.
ORIGIN.md: MIT source attribution. This checkout is independent of the original personal gateway; no shared code mounts, credentials or data.
Release status: scrapling-20261108-223033-e9c05b2 is published and deployed to the VPS (app/control and worker, retaining compatible egress). Local remains on its existing development images. Published image digests: INSTALL.md; current local/VPS references and verification limits: LIVE-STATUS.md.
Runtime
Component | Source / deployment |
MCP frontend, control and one-time credential setup | Shared app image built from |
Chromium/Camoufox and document execution | Worker image built from |
Restricted web-access proxy | Egress image built from |
Dedicated search | Pinned upstream SearXNG image |
Private metadata and artifacts | Pinned upstream Redis and SeaweedFS images |
compose.deploy.yml is the default deployment entry point. scripts/setup_server.py initializes private configuration and operates the stack using published images from malgamoe/mcps; no host application virtualenv or application systemd unit is needed. Only MCP publishes a port. Browsers receive no storage credentials or storage-network access; no application has Docker access.
Follow INSTALL.md for a fresh administrator-owned deployment. Both the local installation and VPS use this layout. Initialization refuses existing configuration; it is not an in-place migration command.
Related MCP server: SDW_Search
Tools and behavior
Usual workflow: search → fetch selected sources → retrieve saved passages; use interactive browsing when needed. Successful calls do not certify extraction completeness or fidelity.
Search: explicit SearXNG/DDGS provider, automatic fallback, domains, page, freshness and language; optional diagnostics. Upstream pagination may overlap; language/freshness are not guarantees and missing publication dates are not inferred from snippets.
Reading:
lightHTTP/Trafilatura;standardScrapling Chromium/structured Markdown;browserScrapling Chromium or Camoufox;autofallback. Stealth requiresstrategy=browserandbrowser_profile=stealth; it does not guarantee access past blocking.Documents: PDF, DOCX, PPTX, TXT, Markdown and HTML text conversion; saved originals and extracted text. Text extraction is not complete document/layout understanding.
Retrieval: bounded cached passages, continuation offsets, an optional heading index, and
web_document_findby literal phrase or ATX Markdown heading. Original evidence remains unchanged.Batch/research: compact source summaries and separate failures. Full text is retrieved separately, even if an older client requests full Markdown in a batch.
RSS: preview, subscribe/unsubscribe/list, fetch, cached entries and read-state updates. Unsubscribe preserves cached entries/read states; retrieve them by feed URL or entry ID. Explicit subscribe/fetch registers the feed again, but an in-flight fetch cannot undo a later unsubscribe.
Browser: open/observe/navigate/click/type/press/scroll/wait/text/find/close/list, bounded text/element waits, explicit back/forward, and native image screenshots with observation-preserving failure warnings. Four ephemeral sessions per owner worker; failures/restarts can expire them.
Live MCP uses one text representation, not duplicate text/structured JSON. Documents default to 12,000-character passages, at most 30,000 per retrieval. Backend JSON is limited to 2 MiB; passage lookup currently retrieves the whole cached document before slicing, so this ceiling still applies. Inline artifacts are limited to 256 KiB; screenshots to 1 MiB. Batches accept 15 URLs and a concurrency argument of 1–10; actual isolated fetch execution has two slots. These are limits, not throughput promises.
Document IDs are cache handles, not archival guarantees: refreshing may replace cached content. URL lookup selects the last saved document; page cache reuse depends on strategy/profile.
Development and operations
uv sync --locked --no-dev
.venv/bin/python -m unittest discover -s tests -qOffline tests exercise the production text-only MCP format, search → batch research orchestration, extraction/presentation regressions, storage maintenance and runtime safety boundaries. Tests include protocol calls, subprocesses and local Unix sockets, but do not start the storage stack or real browsers. Explicit live checks are under scripts/check_*.py; some open/close browser sessions and fetch public sources. Run them only in a coordinated test window, not during someone else's investigation.
The published release passed fresh-image packaged/browser/isolation/cancellation checks and anonymous registry verification. The VPS rolling update passed 56 live MCP calls, search, four reader modes, cached retrieval and Chromium/Camoufox features, plus public HTTPS authentication. Existing records/object bytes and credentials were checked for preservation before acquisition probes. A complete fresh all-new-image installation and actual-host reboot remain unverified. See LIVE-STATUS.md for exact verification and limits; these checks are not a production soak.
Clients use their configured MCP URL and Authorization: Bearer <token> in protected client settings. Reconnect after schema changes. health checks MCP connectivity, not backend health. Never copy credentials between independent deployments.
Preserve .env*, .secrets/, .state/ evidence and Docker data volumes. Never print secrets or use docker compose down -v as cleanup. Manual cache retention is available through python3 scripts/retention.py (read-only dry run by default); see INSTALL.md for pinning, stopped-writer checks and failure recovery. No automatic retention is enabled. Do not restart workers while users are testing without coordinating the interruption.
License
MIT. Third-party dependencies retain their own licenses.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Multi-engine scholarly research server for search, traversal, full text, and reading lists.
Read the web (page text, DNS, WHOIS, TLS, uptime) + research (Wikipedia, arXiv, GitHub, packages).
Read the web (page text, DNS, WHOIS, TLS, uptime) + research (Wikipedia, arXiv, GitHub, packages).
Academic literature search, retrieval, and private library management on top of OpenAlex.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables private web search and webpage content extraction using a local SearxNG instance, prioritizing user privacy and autonomy.226 PyPI2MIT
- AlicenseNot gradedqualityDmaintenanceA provider-neutral Web Search MCP server and CLI that combines live search, scholarly discovery, verified PDF downloads, URL normalization, multi-provider ranking, secure page fetching, caching, and citation-ready research evidence.MIT
- AlicenseNot gradedqualityCmaintenanceAn extended, self-hosted web research server for MCP-compatible clients that provides tools for web search, page fetching, JavaScript rendering, Internet Archive snapshots, and YouTube transcripts.MIT
- AlicenseAqualityBmaintenanceGives AI agents local web search and page reading with no API keys, layering a SearXNG-backed search cache with browser-based fallbacks and returning clean page text from HTML, PDF, JSON, and archived snapshots. Fetches run in Docker alongside the search engine with SSRF protection, egress filtering, and untrusted-content marking so agents can safely retrieve live web data.2MIT