Multi-Agent Research Sandbox
This MCP sandbox implementation by SJTU SAI Agents provides a multi-layered architecture for web search, document parsing, and PDF analysis through both MCP protocol and REST API interfaces. Built with Python and FastAPI, it integrates Google search capabilities, web content parsing with LLM analysis, and ArXiv paper processing using models like GPT-4 and Qwen. The implementation features a modular design with separate API proxy services, configurable LLM backends, and comprehensive tool management through both MCP clients and direct HTTP calls. It serves developers building AI agents that need reliable web research capabilities, academic paper analysis, and flexible deployment options across different integration patterns.
Composite of vulnerability cleanliness, spec conformance, provenance, stability, and usage signals — scanned and weighted by Cognium. Human and agent signals are tracked separately. Last scanned 2026-09-02.
Scan details: Circle-IR · 2026-09-02 · Appeal
View full trust & usage report →Metadata
- Version
- 1.0.0
- Skill type
- atomic
- Execution layer
- mcp-remote
- Category
- devops-ci
- Source
- PulseMCP
- Repository
- github.com/sjtu-sai-agents/mcp_sandbox
- Author type
- human
- Last scanned
- 2026-09-02
- Updated
- 2026-09-02
Use via MCP
Resolve Multi-Agent Research Sandbox from your agent
Streamable HTTP transport at https://api.skillsregistry.net/mcp. No auth for read tools. Discovery: .well-known/mcp.json.
One command in your shell — Claude Code wires it up and verifies the connection. Run /mcp in any session to confirm.
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp--scope user for --scope project to commit it to .mcp.json.