sidmohan0/tesserack
Compiling strategy guides into reward functions for reinforcement learning. Uses Claude Vision to extract unit tests from game guides, then trains agents with dense, interpretable rewards.
This project helps create AI agents that learn to play video games by translating human-written strategy guides into a step-by-step learning path. It takes a game strategy guide (like a PDF) and generates detailed, structured objectives, rewarding the AI for making progress towards these goals. Game developers or AI researchers can use this to train AI agents more efficiently.
Use this if you want to train a reinforcement learning agent for a game using existing human knowledge from a strategy guide, rather than having it learn purely by trial and error.
Not ideal if your game does not have a comprehensive written strategy guide or if you prefer the agent to learn entirely from scratch without human input.
Stars
33
Forks
4
Language
JavaScript
License
—
Category
Last pushed
Jan 30, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/agents/sidmohan0/tesserack"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.