GitHub Repository

一个无感感知上下文、主动推进事务、只在关键决策点打扰用户的协作 Agent。An ambient proactive agent that handles low-risk work automatically and briefs users only when their judgment is needed.

4 starsPython

Steward – an ambient agent that handles low-risk work

Name: Steward – an ambient agent that handles low-risk work
Availability: InStock
Author: 867762462f

by 867762462f·Mar 2, 2026·3 points·0 comments

Visit Project View on HN

AI Analysis

●●SolidBig BrainShip ItBold Bet

Policy-gated autonomous work beats constant summoning, but execution depth unclear yet.

Strengths

•Policy gate abstraction elegantly separates auto-actionable from escalation-worthy tasks.
•Multi-source perception (GitHub, email, calendar, chat, screen) maps real workflow friction points.
•Audit trail + rollback design shows genuine thought on autonomous system safety.

Weaknesses

•Early prototype with 2 stars—unproven in production; core claim (handles 80% of work) unvalidated.
•No concrete examples of what tasks actually auto-execute; definitions of 'low-risk' and 'reversible' vague.

Post Description

I built Steward because most AI assistants still have to be summoned.

The idea here is different: Steward runs in the background, watches signals from tools like GitHub, email, calendar, chat, and screen context, and tries to move low-risk work forward before the user explicitly asks.

The core mechanism is a policy gate between perception and execution. Low-risk and reversible actions can be handled automatically with an audit trail. Higher-risk or irreversible actions must be escalated for explicit approval. Instead of constant notifications, the system is designed to brief the user periodically on what was done, what is pending, and what actually needs judgment.

Right now it is an early local-first prototype. It runs with a simple `make start`, opens a local dashboard, and uses an OpenAI-compatible API key.

I’d love feedback on a few things:

* whether “policy-gated autonomy” is the right abstraction for this kind of agent * where the boundary should be between silent automation and interruption * how people would structure connectors and context aggregation for a system like this