Open sourcePython · ≈81k stars · MIT
Learn Claude Code
Bash is all you need — a nano agent harness built from 0 to 1.
A nano, Claude Code–like agent harness built from scratch in the open as a teaching artifact — every piece of the loop, tools and context handling implemented step by step so you can read a complete harness in an afternoon.
Pros & cons
Each point cites its sources and says what kind of evidence backs it. A second reviewer tried to refute every claim before it was kept.
Pros
- DocsTeaches harness engineering in 17 runnable Python lessons, from the agent loop through tools, permissions, hooks, memory, MCP and agent teams.[1]
- User reportsA HN user called it excellent at explaining, layer by layer, how a Claude Code-like coding agent works.[1]
- MetadataMIT licensed, with English, Chinese and Japanese text in each chapter, about 79.5k stars and only 33 open issues.[1]
Cons
- DocsIt is a course, not a tool: per its 'Course Boundary' section, each chapter isolates one mechanism, and the s15 runtime is assembled for study.[1]
- DocsChapters were renumbered between the legacy 12-lesson track and the current 17-lesson track, and the README warns readers not to mix numbers across them.[1]
- MetadataThere are no tagged GitHub releases, so you cannot pin a version of the material.[1]
- DocsIts framing is openly partisan: it calls Claude Code the most elegant harness and dismisses workflow frameworks as 'Rube Goldberg machines'.[1]
- Best for
- Engineers who want to understand how a coding-agent harness works inside by building one step by step.
- Watch out for
- This is learning material; for a working agent use a maintained harness and treat these lessons as reference.
Assessed Oct 29, 2026 · adversarially reviewedSuggest a correction
People
Founder, shareAI-lab
Claude Code
Activity
- Last push
- · 6d ago
- Latest release
- None published on GitHub
- Last 90 days
- 0 releases
Checked against GitHub on Nov 21, 2026. Release notes are shortened to plain text.
Neighbours on the workbench
Shell · ≈329k stars · MIT
mattpocock/skills
Small, composable agent skills for real engineering work.
Maintained by Matt Pocock (AI Hero)
MDX · ≈323k stars · MIT
obra/superpowers
An agentic skills framework and software methodology that works.
Maintained by Jesse Vincent (obra)
JavaScript · ≈308k stars · MIT
affaan-m/ECC
Agent harness performance optimization system: skills, instincts, memory and security for coding agents.
Maintained by Affaan Mustafa
Connections
Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.
Maintained by · 1
- shareAI-labOrganisation
