Multi-tenant · Secure · Observable

Aggregate tools. Run agents.

BridgeMCP bundles upstream MCP servers into secure virtual endpoints and runs AI agents on top of them that reason, call tools, execute jobs and respond to real-world events.

Get started Log in
zsh — bridgemcp
$ claude mcp add eng-tools \
https://bridge.acme.dev/v/eng-tools/mcp
Added remote MCP server eng-tools · 47 tools
$ bridgemcp agents deploy support-triage.zip
Deployed agent support-triage · 3 skills
triggers webhook:stripe · cron 0 9 * * 1 ● live
One platform, two halves

The gateway and the agents on it

MCP aggregation gateway

Bundle upstream MCP servers into Virtual MCPs: curated capability sets behind one endpoint, with fine-grained access control, per-user OAuth delegation and real-time observability.

Virtual MCPs with one URL per team or workflow
Per-tool allow lists, org scoping, API keys
OAuth per user, tokens never shared
Explore the gateway

AI agent runtime

Deploy agents as markdown bundles that reason, call approved tools, run async jobs and react to schedules and webhooks, automating support, incidents, reporting and operations.

Agents defined in AGENTS.md + SKILL.md bundles
Async jobs with status, results and callbacks
Schedules + Stripe and GitHub webhooks
Explore agents
How it works

From connectors to autonomous work

01

Connect upstreams

Add built-in, custom or community connectors. Configure transport, headers and OAuth clients once.

02

Bundle into Virtual MCPs

Pick servers, filter tools and set access rules. Each bundle becomes a virtual endpoint.

03

Point agents at it

Your own agents or BridgeMCP-hosted ones. One URL and one credential each.

04

Automate with triggers

Run agent jobs on schedules or webhooks. Track state, tool usage and outputs.