autocontext

A recursive self-improving harness for agent improvement.

AI AgentsUnknownOpen source
Visit tool

github.com

About autocontext

autocontext is a harness for agent improvement. It runs tasks against evaluation, keeps useful lessons, discards dead ends, and leaves traces, reports, playbooks, datasets, and optional local-model training artifacts for the next run.

Description summarised by AI from the sources listed below.

Key features

  • recursive self-improving harness
  • task evaluation
  • useful lesson keeping
  • dead end discarding
  • traces, reports, playbooks, datasets, and local-model training artifacts

Use cases

  • agent improvement
  • task automation

Pricing

Pricing model: Unknown — we have not been able to confirm pricing from the official website, so nothing is stated here.

Pricing summarised by AI from the sources listed below.