ondraulehla/agent-flight-recorder

#AI agent

Flight recorder for AI agents: run tasks N times in isolated E2B sandboxes, record full traces, measure pass rate, flakiness and cost, and diff trajectories

Visit GitHub

Python

2 contributors

MIT License

website

Social mentions

Recent discussions about this repository across the web

Practical consequence: if you benchmark agent cost with parallel runs, the first wave inflates your average. Warm the cache, report the median, or drop the wave. For this task it moves the number…

@ondrejulehla · x.com

So I write the assertion before the prompt now. Test-first for agents: the prompt is the implementation, the assertion is the spec. Fixed the assertion to match what "sorted" actually means, reran:…

@ondrejulehla · x.com

So I write the assertion before the prompt now. Test-first for agents: the prompt is the implementation, the assertion is the spec. Fixed the assertion to match what "sorted" actually means, reran:…

@ondrejulehla · x.com

Repository activities

repository's daily and monthly activities across stars, forks, merged PRs, issues, and closed issues