The experience graph for AI agents. Self-hosted memory, human-reviewed decisions, and outcome-backed datasets for model evaluation and learning.
-
Updated
Oct 4, 2026 - TypeScript
The experience graph for AI agents. Self-hosted memory, human-reviewed decisions, and outcome-backed datasets for model evaluation and learning.
[arXiv] ExpHarness: Model-Agnostic Experience Learning through a Trainable Harness
A local experience graph for coding agents. Connect code, Codex and Claude Code sessions, decisions, failures, and fixes.
Local-first compounding intelligence for AI agents. Experience Graph, Memory Bank, Growth Merge, learned skills. MCP-native for any model. MIT.
To associate your repository with the experience-graph topic, visit your repo's landing page and select "manage topics."