greyhaven-ai/autocontext

AutoContext uses LLM-based judging with rubric-driven scoring to evaluate and improve agent outputs. Use it to run improvement loops, queue background evaluations, check run status, inspect artifacts and session lineage, or discover scenarios.

Autocontext is a control plane for evaluating agent behavior, preserving run artifacts, exporting training data, and distilling stable behavior into local runtimes. Use it when a Hermes agent needs to run scenarios, inspect Hermes Curator state, or prepare local MLX/CUDA training data through the autoctx CLI.

Autocontext Consumer reads and moves knowledge produced by Autocontext, including playbooks, lessons, and hints. Use it to check what has been learned, read current playbooks, and export or import knowledge between checkouts.

Autocontext Creator runs an Autocontext improvement loop over a task and writes what it learned to disk. Use it when an agent needs to create knowledge through scenarios, judge or improve a single output, or inspect what a run produced.

Grid CTF Ops provides operational knowledge for the grid_ctf scenario, including strategy playbook, lessons learned, and resource references. Use when generating, evaluating, coaching, or debugging grid_ctf strategies.