agentgate

agentkitai/agentgate

Human-in-the-loop approval workflows for AI agents

PersonalEnterpriseAI & LLM toolingMCP serversSecurity & identityWebDocker / self-hostCLILibrary / SDKPermissive

At a glance

stars on GitHub
33
contributors
2+
last commit
latest release, 4 months ago
v0.15.0
license
MIT

About

AgentGate is an approval gateway and policy engine for AI agents. It evaluates proposed agent actions against configurable rules to auto-approve safe tasks or route sensitive requests to humans through a web dashboard, Slack, Discord, or email. The stack can be self-hosted with Docker Compose or run locally using Node.js.

From the repository: “Approval workflows for AI agents”

Latest release: agentgate v0.15.0 — revocable ingest keys + governance docs,

Source code only: this release has no files to download.

agentgate and its alternatives: GitHub stars, last commit, license and worth score
ProjectStarsLast commitLicenseWorth
agentgateThis project
33MIT50
mcp-approvalsHuman-in-the-loop approval gate and audit log for AI agents
0MIT28
agentgateSelf-hosted API gateway for AI agents with write approval
45 (best)Unknown37
portunusBoundary-only secret broker keeping credentials out of LLM contexts
2MIT53 (best)
mcp-wardenLocal-first credential broker and MCP server for AI agents
9Apache-2.050

▲ marks the most stars and the best worth score.

agentkitai.github.io
Preview of agentgate
Preview of agentgate at agentkitai.github.io

Topics