LearnHarness Launches AI Agent Harness That Tests Understanding Before Teaching

LearnHarness released an AI agent harness that assesses a learner's existing understanding before planning and teaching lessons toward a goal. The tool builds a persistent knowledge graph across sessions, supports macOS, Windows and Linux installers, and works with user-supplied API keys for more than 30 model services with no subscription fee.
Event version 1.
Key insights
Assessment before instruction
Before teaching, the agent tests where the learner's understanding ends on each part the lesson depends on. Answers, confidence and reasoning determine the starting point.
Dependency-based lesson planning
The agent works out a dependency map from the starting point to the learner's goal and arranges a learning route along it.
Stepwise teaching with correction
Lessons advance one reasoning step at a time: the learner guesses first, then receives an explanation, then is checked. Misunderstandings are corrected before further material builds on them.
Persistent knowledge graph
Learned content is stored as a knowledge graph that carries across sessions. Notes on reasoning and uncertain points inform what the agent teaches next.
Cross-platform, bring-your-own-key
Installers are available for macOS, Windows and Linux. Users enter their own model API key, with support for more than 30 model services, no subscription fee, and real-time cost display.