Add Orihime — cross-repo code knowledge graph MCP server for Java/Kotlin/JS/TS

[srinivasan-sundaresan95]

What is Orihime?

Orihime is an open-source cross-repository code knowledge graph that indexes Java, Kotlin, JavaScript, and TypeScript codebases into an embedded KuzuDB graph using tree-sitter, and exposes 30+ MCP tools for AI assistants.

What it enables

• Call-flow tracing — trace any endpoint through service → repository → upstream API chains across multiple repos
• Taint analysis — multi-hop BFS value-flow from @RequestParam/req.body to SQL/exec/SSRF sinks; entry-point reachability filtering reduces false positives 30–50%
• Security reports — OWASP Top 10, CWE, PCI DSS, STIG frameworks; second-order injection detection; custom sources/sinks via YAML
• Performance analysis — Gatling/JMeter ingestion; Little's Law capacity per endpoint; O(n²) and N+1 detection
• License compliance — scan Maven/Gradle dependencies against SPDX; flag GPL/AGPL/LGPL
• Cross-service taint — taint paths that cross REST service boundaries via HTTP clients

Why it's a good fit for this list

• MIT licensed, fully open-source
• Stdio MCP transport (works with Claude Code, Claude Desktop, any MCP client)
• Embedded DB — no daemon, no server infrastructure required
• pip install orihime — pypi.org/project/orihime
• Benchmarked: 95% fewer tokens and 200–1400× faster queries vs source-file reading baseline on an 845-file Kotlin + 224-file Java codebase

[srinivasan-sundaresan95]

Closing — submitting to the MCP Server Registry at github.com/modelcontextprotocol/registry instead as the bot suggested.