"Stop repeating yourself to every new AI agent."

ACSS is a lightweight, portable protocol that captures the intent, decisions, and context of a coding session. It bridges the gap between human developers and AI assistants (or between two AI assistants), preventing "Context Rot" when switching tools.

I have included a full demo video

https://youtu.be/GgGUTj-WXK8

This script simulates a full development lifecycle:

1. Project Initialization: Auto-detects tech stack.
2. Context Logging: captures decisions & errors.
3. Chat Import: Ingests ChatGPT/Claude conversations.
4. Instant Handoff: Transfers context between AIs.
5. Watch Mode: Real-time file tracking.

ACSS captures your development flow by combining:

1. Complete Project Analysis: Directory structure, file tree, and tech stack detection.
2. AI Conversation History: Imports chats from ChatGPT, Claude, etc.
3. Real-time Context: Watches file changes and git commits as you work.

The output is a structured, AI-optimized context (.acss/session.acss.json) that can be handed to any AI coding assistant.

• Node.js (v16+)
• npm (v8+)
• Git

Clone the repository and install dependencies:

git clone https://github.com/sp25126/ACSS_context.git
cd ACSS_context

# Install Core Logic
cd packages/core
npm install
npm run build

# Install CLI Tool
cd ../../cli
npm install
npm run build
npm link

Verify the installation by running:

acss --help

Go to any project directory and run:

acss init
# ACSS will scan your project and create a session file.

Export your conversation from ChatGPT or Claude and import it:

acss import ./chat-export.json

Create a context-rich prompt for your next AI assistant:

acss load --for claude > handoff_claude.txt

Automatically track file changes and git commits:

acss watch

Your session state is stored in .acss/session.acss.json. It includes:

• Project Metadata: fileTree, techStack, totalFiles.
• Current Task: intent, status.
• History: decisions (manual + git), errorsEncountered.
• Context: chatContext (imported insights), filesModified (watch logs).

I am currently looking for Job / Internship opportunities.

If you are interested in my work or would like to discuss potential collaborations, please reach out!

📧 Email: saumyavishwam@gmail.com 🔗 GitHub: https://github.com/sp25126

This project is licensed under the MIT License - see the LICENSE file for details.