asimons81

asimons81 / hermes-dreaming

Public

Staged self-improvement engine for Hermes-style memory, skill, and fact updates with explicit review and apply/discard gates.

ai-agents cli mcp memory open-source
27
0
89% credibility
Found May 26, 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

Hermes Dreaming is a tool that helps you manage what your AI assistant knows about you. You write simple notes describing updates you want—like new preferences, facts, or skills—and the tool stages these changes for your review. Nothing is changed automatically; you always get to see what's proposed, approve what you want, and the tool safely writes the updates while keeping backups. It's designed for people who want control over their AI's memory without writing code.

How It Works

1
💭 You want your AI to remember things better

You've been using an AI assistant and wish you could easily update what it knows about you, your preferences, and the skills it should have.

2
🔌 You install the plugin

You add this tool to your AI assistant setup with a simple command. Everything is ready to go.

3
✍️ You write down what you want changed

In your notes or documents, you write special markers describing updates—like 'remember that I prefer short responses' or 'add a new review skill'.

4
🔍 The tool scans your notes

You ask the tool to look through your documents. It finds all your markers and prepares a list of suggested changes.

5
📋 You review what was proposed

Before anything changes, you get to see exactly what edits the tool wants to make. You can check each one carefully.

6
You decide what to keep
✅
You approve and apply

The tool safely writes the changes to your AI's memory, keeping a backup in case you need to undo later.

🗑️
You discard instead

You archive the proposals without making any changes. Nothing in your AI is affected.

🎉 Your AI is now updated

Your assistant now has your latest preferences, memories, and skills—all because you wrote a few notes and clicked apply.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 27 to 27 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 hermes-dreaming?

hermes-dreaming is a staged self-improvement engine for AI agent memory, skills, and facts. It scans source files for explicit update instructions, stages proposed changes in reviewable artifacts, and only writes to live state after you explicitly approve them. Built in Python with a CLI tool called `dreaming`, it supports both offline marker-based analysis and OpenAI-compatible LLM providers. The workflow centers on create, diff, validate, apply, and discard commands that give you full control over what gets committed to your agent's knowledge base.

Why is it gaining traction?

The staged review workflow is the hook. Instead of letting an AI agent silently modify its own memory, hermes-dreaming forces an explicit apply/discard gate before any changes touch live state. This appeals to developers who want audit trails and human oversight. The offline marker provider (parsing DREAM: lines in source files) offers a lightweight path without API costs, while the OpenAI-compatible provider enables LLM-powered analysis for more complex scenarios.

Who should use this?

Developers building Hermes-style AI agents that need to update their own memory and skills over time. Teams maintaining shared agent configurations where multiple contributors propose updates. Anyone tired of AI agents silently modifying their own state without review.

Verdict

hermes-dreaming solves a real problem for AI agent development with a clean, reviewable workflow. The credibility score is 0.8999999761581421%, reflecting its early stage with modest adoption. At 27 stars, it's nascent but functional, with solid documentation and basic test coverage. Worth trying if you're building Hermes-style agents or need a safe way to manage agent self-improvement.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.