A guard layer that checks agent tool calls and screens what tools send back
Emerged finds new products worth a look, daily — what’s new today
If an agent in your stack can run shell commands or spend money, this gives you one place to say no before it acts and a record of everything it asked for.
AgentTrust sits between an autonomous agent and the tools it calls. Each proposed action is inspected for prompt injection and leaked secrets, scored for how much damage it could do, and checked against capability policies; risky ones wait for a person to approve them, and every decision is written to a tamper-evident audit trail. You clone it, run npm install, build the client and start the server, and the dashboard comes up on localhost:4000. SDKs ship for Python, TypeScript, Go and Rust, with a Dockerfile and compose file for self-hosting. MIT, 42 commits, no tagged release yet.
More like this
Approval gates and an audit ledger for coding agents left running unattended
A local ledger of every tool call your coding agents make
An open-source safety layer that checks every AI agent action before it runs
Self-hosted AI coworkers, each with its own computer and a policy gateway
Inventory every AI agent on your machine and what it can reach