Social Media Agent

Agent workflow experiment for turning URLs or source material into social posts, with LangGraph/LangChain orchestration and several optional integrations.

Project Status

This repository is being open sourced as-is. It may or may not be in working condition. Most of the work here came from vibe-coding experiments by an engineer with very little coding experience, so expect integration churn, environment drift, and setup gaps.

Main Pieces

• src/: TypeScript/LangGraph agent graph
• memory-v2/: optional Python memory graph
• slack-messaging/: Slack-related helpers
• docs/, reports/, scripts/: supporting materials

Dependencies

• Node.js 20+
• Yarn 1.x
• Python 3.11+ if you plan to use the Python memory components
• API keys/accounts for the integrations you actually enable

Local Run

Install the JavaScript dependencies:

yarn install

If you want the optional Python memory graph:

cd memory-v2
python -m venv .venv
. .venv/bin/activate
pip install -e .

Copy one of the example environment files and fill in the values you need:

cp .env.quickstart.example .env

Start the in-memory LangGraph dev server:

yarn dev

Other useful commands:

yarn build
yarn test
yarn test:int

Notes

• This repo has a lot of optional integrations. A local boot is most realistic if you start in the quickstart mode and only enable the minimum services.
• Historical deployment artifacts were removed as part of the open-source cleanup.