Launching today

MELRA

Launching today

Safer MCP tool execution with approvals and recovery

2 followers

Safer MCP tool execution with approvals and recovery

2 followers

Visit website

An agent can finish a write even when its response gets lost. Retrying may duplicate the change. MELRA is an open-source local runtime for MCP and CLI tools. It checks policy and scoped approvals before actions, verifies declared outcomes, and keeps a redacted record. If a write is interrupted and its result is unclear, MELRA stops for reconciliation instead of replaying it. Alpha software; requires Node.js 22.13+.

Free

Launch tags:Open Source•Developer Tools•Artificial Intelligence

Launch Team

Maker

📌

I’m building MELRA at XAGI Labs. The current alpha runs locally and requires Node.js 22.13+. Try it in a throwaway workspace with `npx @melra/cli@alpha setup`. It asks before writing an MCP client config. I’d especially value a run through one client flow and feedback on what happens when a tool call is interrupted. Partial or failed runs are useful too. What would you need to see before trusting an agent with local tool access?

Report

2d ago

Forum Threads

p/melra

•

17h ago

What do you do when a tool call times out after a write?

I m building MELRA, a local runtime for MCP and CLI tools, and I m testing this recovery edge case: a write may reach the provider even though the agent never gets a response. Retrying can duplicate it, but waiting without a way to reconcile is frustrating. In the current alpha, MELRA parks that attempt until state can be checked. What evidence would you trust before allowing a retry, especially when provider lookups are slow?

View all