BlinkDL

BlinkDL / Agen

Public

Agen is a minimalist language for agent loops and state machines.

20
0
100% credibility
Found Mar 30, 2026 at 27 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Agen provides a symbolic language and Python runtime for defining and running state-driven loops that simulate behaviors like NPC routines or AI agents with tool interactions.

How It Works

1
🔍 Discover Agen

You hear about Agen, a simple way to make characters act out stories or smart helpers follow step-by-step plans using easy symbols.

2
📖 Explore Examples

You look at ready-made stories like a shopkeeper's daily routine or an AI helper that checks folders and updates lists.

3
▶️ Run Your First Story

You start the shopkeeper example and watch her wake up, open her stall, and head home as the log fills with her day's events.

4
✏️ Tweak the Rules

You change a few symbols to add new actions, like extra tasks or different locations, feeling like directing a play.

5
🛠️ Try Smart Helper

You launch the advanced helper that reads notes, edits files, manages to-dos, and chats back with updates.

6
🔄 Test and Adjust

You run it again, see the results, and fine-tune until the behavior matches exactly what you imagined.

🎉 Your Creation Lives

Your custom character or helper runs smoothly, producing logs and outputs that bring your ideas to life effortlessly.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 27 to 20 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 Agen?

Agen lets you define agent loops and state machines in a minimalist Python-based language, using indentation and UTF-8 symbols for rules like `(condition)➜ actions`. It runs your agenda-driven logic in a tight loop, matching the first rule on state variables, executing blocks sequentially, and restarting until no matches or a step limit hits. Developers get a declarative way to model agent behaviors—like NPC routines or LLM tool chains—without verbose Python if-else spaghetti.

Why is it gaining traction?

Unlike bloated FSM libraries or raw Python loops, Agen's agent provocateur syntax shines for AI workflows: slots bind values cleanly, templates handle dynamic strings, and helpers plug in functions like model queries or bash commands. It compiles to optimized loops, making agent zeta prototypes fast to iterate. Early adopters dig the AI-coded feel, with a 1k dataset on Hugging Face for training models on Agen scripts.

Who should use this?

AI engineers building ReAct-style agents with tool loops, game devs scripting simple NPC state machines, or backend folks modeling minimalist workflows like job queues at agentur für arbeit jobbörse. Skip if you need complex graphs—it's for linear, rule-matching agendas.

Verdict

With 20 stars and 1.0% credibility, Agen is raw and experimental—docs are README-only, no tests—but its python state machine purity hooks for quick agent prototypes. Try for toy agents; production needs polish.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.