sagents-ai

sagents-ai / sagents

Public

Build interactive AI agents in Elixir with OTP supervision, middleware composition, human-in-the-loop approvals, sub-agent delegation, and real-time Phoenix LiveView integration. Built on LangChain.

agent agentic ai elixir hitl
148
13
100% credibility
Found Feb 12, 2026 at 25 stars 6x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Elixir
AI Summary

Sagents is an Elixir library for building interactive AI agents in Phoenix apps with real-time updates, human oversight, sub-agents, virtual filesystems, and conversation persistence.

How It Works

1
🔍 Discover Sagents

You hear about Sagents, a tool that helps create smart chat assistants for your web app that think step-by-step with your guidance.

2
📦 Add to your app

Simply tell your app to include Sagents, like adding a new ingredient to a recipe.

3
🧠 Connect AI brains

Link popular AI services so your assistant can understand and respond like a helpful friend.

4
✨ Build your first assistant

Use a magic command to create a full chat helper with safety checks, file handling, and task lists in seconds.

5
💬 Chat and watch it work

Start talking to your assistant in real-time; it shows its thinking, asks for okay on big steps, and handles files.

6
🛡️ Guide with approvals

When it wants to do something important like edit files, you review and say yes, tweak, or no.

🎉 Your smart helper lives!

Your web app now has a powerful, safe AI assistant that chats, plans, and gets work done just like you wanted.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 148 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is sagents?

Sagents lets Elixir developers build interactive AI agents for Phoenix LiveView apps, handling real-time conversations with OTP supervision, PubSub events, and Presence for smart shutdowns. It adds human-in-the-loop approvals for risky actions like file ops, sub-agent delegation for parallel tasks, virtual filesystems, and state persistence via database generators. Built on LangChain, it turns LLM calls into scalable, concurrent agents for building interactive dashboards or chat prototypes.

Why is it gaining traction?

Unlike plain LangChain for batch jobs, sagents delivers production orchestration with middleware for todos, summarization, and delegation—plug in capabilities without custom loops. Generators scaffold persistence schemas, LiveView handlers, and coordinators in one command, slashing setup for real-time UIs. Early adopters dig the resource efficiency: idle agents auto-shut down when viewers leave.

Who should use this?

Phoenix devs crafting live chat apps or interactive websites where AI needs oversight, like code reviewers approving file changes or behavioral health agents flagging sensitive responses. Perfect for solo makers building github projects with AI demos, or teams prototyping multi-user agents behavioral health tools or github copilot-style assistants.

Verdict

Solid for Elixir shops eyeing agent frameworks—generators and demos make it dead simple to build interactive prototypes fast, despite 19 stars and 1.0% credibility signaling early maturity. Skip if you need proven scale; otherwise, mix deps.get and iterate.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.