ravi3594444/swarmtrace — reverse-engineered prompt

Build me a small Python library called SwarmTrace. I want it to feel like pytest for AI agents, so someone can write simple tests for a multi agent workflow and see what each agent did, what tools it called, what it returned, how long it took, and where things failed.

Please make a clean package with an easy API like creating an Agent, running it with a prompt, and getting back a result object with output, latency, trace steps, and any error. Add a SwarmTest or similar helper for grouping agent runs and checking regressions. It should work with mocked agents by default, but leave simple hooks for OpenAI, Anthropic, LangChain, or local LLM calls later.

Include pytest examples, basic tests, a README with install and quick start instructions, and make it ready to publish as a Python 3.10+ package. Keep it practical and hackathon friendly. Look up current docs online if you need to.