mcp · Global

navra

Secure MCP gateway for AI agents — deny-wins ACLs, IFC, in-process ML safety, 138 Kani proofs

Feature or claim this listing

Represent navra? Send a claim for review so we can discuss listing updates. Claims do not automatically grant ownership or verification.

Data offered docs-verified: MCP gateway · Information Flow Control · deny-wins ACLs · credential brokering · audit blackbox · agent SDK · multi-agent flow orchestration
Formats docs-verified: MCP streamable HTTP · MCP stdio · MCP WebSocket · MCP Unix socket
Access docs-verified: Free and open source under Apache-2.0; self-hosted, built from source with cargo

Use cases llm-inferred: Put a permission-controlled gateway in front of MCP tools used by agents on a client engagement, so file and domain access is denied by default · Keep a tamper-evident audit log of AI agent tool calls for engagement QA records · Expose an internal OpenAPI-described environmental or asset API to agents as MCP tools, with OAuth
Audience llm-inferred: AI platform engineers · security and IT engineers · Rust developers · IT governance teams

Constraints llm-inferred: Not a data source: no water, civils, GIS, hydrology, BIM or environmental datasets are described in DOCS · Catalogue category civils_bim is not supported by DOCS · No geographic or temporal coverage is described; ENTRY region 'Global' is metadata only · Requires a Rust toolchain and building from source (Quickstart) · The '138 Kani proofs' count in the ENTRY description does not appear in DOCS · Much of the feature set (ML safety filters, model serving, voice, vision) is irrelevant to data-retrieval use

Auto-extracted, unverified fields: tools, auth. Check upstream documentation.

Tools/install/auth auto-extracted from README; quality score heuristic (stars, recency, tool count, licence).

Metadata observed 16/10/2026 · research sweep: github-search