laude-institute/headlong
Headlong is an open-source agent microharness written in Bash that provides persistent agency and recursive LLM integration.

Headlong is a Bash-based agent harness of roughly 10K lines that runs an agent in a self-guided loop inspired by human inner monologue. It integrates with chat platforms such as Slack and Telegram, routing all messages into a single thought stream that the agent processes continuously. At its core is shellm, a Bash implementation of a recursive language model that lets the agent invoke further LLM-driven reasoning from within its own loop.
Frequently asked
- What is laude-institute/headlong?
- Headlong is an open-source agent microharness written in Bash that provides persistent agency and recursive LLM integration.
- Is headlong open source?
- Yes — laude-institute/headlong is open source, released under the Apache-2.0 license.
- What language is headlong written in?
- laude-institute/headlong is primarily written in Shell.
- How popular is headlong?
- laude-institute/headlong has 1.1k stars on GitHub.
- Where can I find headlong?
- laude-institute/headlong is on GitHub at https://github.com/laude-institute/headlong.