Hi HN — I built Elf0, a command-line tool to define and run AI agent workflows in YAML. It helps you iterate on small multi-step "agents" without scaffolding a whole codebase.
Why: I keep bumping into tasks where a single prompt isn’t enough (e.g., extracting quote data from an insurance PDF). Defining the workflow in YAML makes it easy to version prompts, parameters and logic, and to feed the whole spec back to an LLM to improve.
What it does:
- Run single- or multi-step agents from a YAML spec.
- Reference files or directories in prompts with @file / @dir.
- Improve a workflow with a built-in command.
- Use OpenAI or Anthropic APIs, or run locally via Ollama.
- Optional MCP server example (YouTube transcript analyser).
- Interactive mode for step-by-step runs.
Anyway it's been a lot of fun and really useful, let me know any ideas you might have to improve it?
The agent patterns described in Anthropic's article was an inspiration: https://www.anthropic.com/engineering/building-effective-age... I then used Nvidia's AgentIQ YAML spec as inspiration.
Why: I keep bumping into tasks where a single prompt isn’t enough (e.g., extracting quote data from an insurance PDF). Defining the workflow in YAML makes it easy to version prompts, parameters and logic, and to feed the whole spec back to an LLM to improve.
What it does: - Run single- or multi-step agents from a YAML spec. - Reference files or directories in prompts with @file / @dir. - Improve a workflow with a built-in command. - Use OpenAI or Anthropic APIs, or run locally via Ollama. - Optional MCP server example (YouTube transcript analyser). - Interactive mode for step-by-step runs.
Anyway it's been a lot of fun and really useful, let me know any ideas you might have to improve it?