CacheBash
Coordinate multiple AI coding sessions with shared messaging, tasks, session state, and alerts through one MCP endpoint.
Overview
CacheBash MCP is an MCP server for the Build phase that lets AI sessions exchange messages, tasks, session data, and alerts over authenticated streamable HTTP.
What is this MCP server?
- Remote streamable-http MCP on Google Cloud Run
- Bearer API key required via Authorization header
- Cross-session messaging so agents can hand off context
- Tasks and session management plus alert hooks
- Version 2.0.0 hosted service—no local server process to run
- Remote streamable-http MCP version 2.0.0
- Hosted at cachebash-mcp-922749444863.us-central1.run.app/v1/mcp
- Required Bearer Authorization API key
What problem does it solve?
Parallel agent tabs do not share state, so you re-explain context and lose track of which session owns which task.
Who is it for?
Builders juggling multiple MCP-enabled clients who want a lightweight coordination layer without building custom webhooks.
Skip if: Single-threaded workflows, air-gapped environments, or teams unwilling to store session metadata on a third-party hosted API.
What do I get? / Deliverables
Sessions message each other, track tasks, and surface alerts so one agent can continue where another left off.
- Cross-session messages visible to connected agents
- Shared task and session records across MCP clients
- Alert signals when sessions need attention or handoff
Recommended MCP Servers
Journey fit
Complex builds split across parallel agent threads; tooling that links sessions belongs where you assemble agent workflows, not only in post-launch ops. Agent-tooling is the shelf because CacheBash is infrastructure for how sessions talk to each other, not a single feature generator.
How it compares
Multi-session agent coordination MCP, not a code indexer or git integration.
Common Questions / FAQ
Who is CacheBash MCP for?
Indie builders and power users running more than one AI agent session who need messaging, tasks, and alerts in one place.
When should I use CacheBash MCP?
Use it during active Build work when you split frontend, backend, and research across sessions that should stay in sync.
How do I add CacheBash MCP to my agent?
Configure the remote streamable-http URL https://cachebash-mcp-922749444863.us-central1.run.app/v1/mcp with Authorization: Bearer YOUR_API_KEY in your MCP client headers.