- Home
- Guides and documents
- Single agent loop kit
Single agent loop kit
Kit for folding subagents into one skill-loading agent loop
What it is
Starter kit and runnable example (a support desk with a router and six subagents) for measuring a multi-agent system, merging subagents that always run together and converting the rest into skills and tools, scoring each change on the same eval.
Good for
- Evaluate whether to drop subagents
- Convert a subagent to a skill
- Run the support-desk example
What you need
Python and an Anthropic API key for the example
Worth knowingReference implementation, not maintained or accepting contributions.
Sources
- https://github.com/anthropics/single-agent-loop-kit (project README, read 13 Nov 2026)
- https://api.github.com/repos/anthropics/single-agent-loop-kit (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.