What problem does this solve?
AI coding tools (Cursor, Cline, Continue, Claude Code) communicate via MCP (Model Context Protocol). This traffic currently bypasses Routerly entirely, making it invisible to cost tracking, routing policies, and auth.
Proposed solution
MCP-compatible gateway layer that proxies MCP tool calls through Routerly:
- Apply existing routing policies, cost tracking, and auth to MCP traffic
- Dashboard: MCP request volume, latency, cost by tool/agent
- Compatible with MCP clients that support configurable server URLs
- No changes required to MCP tools themselves
Alternatives you've considered
Running a separate MCP proxy alongside Routerly. This duplicates infrastructure and splits observability across two systems.
Who would benefit from this?
Developer teams using AI coding assistants alongside Routerly. As MCP becomes the standard protocol for agentic tool use, this becomes a table-stakes feature for any LLM gateway.
Additional context
Requesty ships MCP Gateway as a dedicated product section. MCP adoption is accelerating rapidly in 2025–2026.
What problem does this solve?
AI coding tools (Cursor, Cline, Continue, Claude Code) communicate via MCP (Model Context Protocol). This traffic currently bypasses Routerly entirely, making it invisible to cost tracking, routing policies, and auth.
Proposed solution
MCP-compatible gateway layer that proxies MCP tool calls through Routerly:
Alternatives you've considered
Running a separate MCP proxy alongside Routerly. This duplicates infrastructure and splits observability across two systems.
Who would benefit from this?
Developer teams using AI coding assistants alongside Routerly. As MCP becomes the standard protocol for agentic tool use, this becomes a table-stakes feature for any LLM gateway.
Additional context
Requesty ships MCP Gateway as a dedicated product section. MCP adoption is accelerating rapidly in 2025–2026.