Back to directory

claude-code-workflow-creator

A Claude Code skill that teaches the assistant to author deterministic multi-agent orchestration scripts for Claude Code's unreleased Workflow tool, where ordinary JavaScript control flow drives fresh-context agent calls.

claude-code-workflow-creator

This is a Claude Code skill package that teaches Claude to write workflows: JavaScript files whose loops, conditionals and branching are ordinary code the developer controls, where only the individual agent calls at the leaves spend model tokens and each runs in its own clean context window. The result is multi-agent work that behaves the same way on every run and can resume if it stops partway through.

The repository is the skill itself, ready to be dropped into a Claude Code skills folder, along with a full reference of every global, option and constant, a library of copy-paste orchestration patterns such as fan-out, pipeline and loop-until-budget, starter templates for the three core shapes, six complete runnable example workflows, and a linter that checks a workflow file against the parser's rules before it runs.

It targets Claude Code's Workflow tool, which is unreleased and hidden behind an environment variable at the time of writing, so the workflows it writes will not run until that feature ships.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.