The AI builder

pip install "tolquane[ai]"
export ANTHROPIC_API_KEY=...         # or OPENAI_API_KEY with --provider openai
tolquane build "read urls.txt, fetch each with 8 workers, write url, status and size to status.csv"

The builder writes one short, commented flow.py in the house style, checks its wiring, runs it on the sync runtime with a sample it makes up or with --sample file, fixes what fails, and then asks what to change. Claude Opus 5 is the default; GPT works with --provider openai. The same loop is a function:

from tolquane.ai import build
result = build("count words per line in log.txt, 8 workers", workdir="out")
print(result.summary)

The same builder is a panel in the GUI: tolquane web, then AI builder beside the editor, with the open flow as its context (see the user guide).

Every flow the builder writes defines build(source=None), which returns the graph and swaps in tq.from_iterable(source) when a sample is given, and main(). That is what tolquane check, run, explain and draw expect from any file.

Ten flows it wrote, unedited, with their transcripts, are in examples/generated; none of the ten conversations had a tool error. Generated code runs on your machine in a subprocess with a timeout. Keys are read from the environment and never stored.