In-depth architectural comparison of the Touch Browser and Onecite MCP servers. Compare execution transports, security boundaries, tool capabilities, quality scores, and ready-to-paste client installation snippets for Claude, Cursor, Windsurf, and VS Code.
At a Glance & Executive Verdict
Touch Browser
Research · Local stdio
Quality: 52/100 (Good) | Auth: No auth required
Onecite
Research · Local stdio
Quality: 44/100 (Fair) | Auth: No auth required
Verdict Summary: Choose Touch Browser if you need specialized Research tools running via a local process. Choose Onecite if your workspace requires Research integration with local subprocess execution. Both servers can be configured concurrently in your client's mcpServers manifest.
Which MCP Server Should You Choose?
Choose Touch Browser when:
You need dedicated capabilities in the Research domain.
You prefer local stdio subprocess transport architecture.
Your security boundary fits: No auth required (Free / Open Source).
Primary tools included: Page-local claim verdicts, Confidence bands and review recommendations, Citation URLs with retrieval timestamps.
You need dedicated capabilities in the Research domain.
You prefer local stdio subprocess transport architecture.
Your security boundary fits: No auth required (Free / Open Source).
Primary tools included: Four-stage parse, identify, enrich, and format pipeline, BibTeX and CSL-JSON output, Metadata completion for authors, journals, volumes, pages, and abstracts.
Evidence-grounded web research for AI agents. Verifies page-local claims with citations, confidence bands, policy reports, and multi-page session synthesis via MCP.
Generate academic citations from DOI, arXiv, titles, or URLs in BibTeX, APA, MLA formats.
Touch Browser is categorized under Research and uses a local stdio subprocess. In contrast, Onecite belongs to Research using local stdio subprocess. Select Touch Browser when you need capabilities focused on research and Onecite when you require tools for research.