Skip to content

Releases: Qitor/qitos

QitOS v0.3.0

08 Apr 14:20
@ravenSanstete ravenSanstete
v0.3.0
3efaab7

Choose a tag to compare

View all tags
QitOS v0.3.0 Latest
Latest

QitOS v0.3.0

QitOS v0.3.0 turns reproducible agent runs into a first-class product feature.

Who this release is for

  • Agent researchers iterating on prompts, parsers, tools, and memory policies
  • Benchmark users who want GAIA, Tau-Bench, and CyBench on the same kernel as their agents
  • Builders who need trajectory inspection, replay, and diff instead of framework glue

What shipped

  • Official reproducible-run contracts with RunSpec, ExperimentSpec, and normalized BenchmarkRunResult
  • New qit bench workflow for run, eval, replay, and export
  • qita replay, export, and diff surfaces for review-grade trajectory analysis
  • A packaged minimal coding-agent demo via qit demo minimal
  • Updated English and Chinese quickstarts, CLI docs, and first-agent guides around the same minimal coding-agent path

Run it in minutes

pip install "qitos[models]"
export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://api.siliconflow.cn/v1/"
export QITOS_MODEL="Qwen/Qwen3-8B"
qit demo minimal
qita board --logdir runs

qit demo minimal seeds a tiny buggy workspace, asks a model-backed coding agent to repair it, verifies the patch, and writes a qita-ready trace under ./runs.

Why it matters

QitOS is not trying to be the widest app framework. The core idea is narrower and more research-friendly: one AgentModule + Engine kernel for method iteration, benchmark execution, and trajectory inspection.

Get involved

If this direction resonates, please star the repo, open an issue, or contribute. Early feedback on benchmark adapters, qita workflows, memory/history policies, and cyber-agent use cases is especially valuable.

Assets 2
Loading