Buzz vs BAND
Bench log · buzz github integration, buzz pull request

Bench: chat to PR on Buzz

Last verified: 2026-08-19

Desk draft · not yet bench-measured

The method below is fixed and published. Timers read Not yet measured until the run completes.

Buzz's stated core loop runs chat to plans to project management to coding to reviews to pull requests. This entry takes one real bug from a public repository through that loop and times each hop. It is the proof behind the Buzz developer workflow score, or its correction. Desk-drafted from the bench brief. Method is fixed and published below. Numbers land here after the run, and until then every timer on this page reads Not yet measured.

Issue to open pull request
Not yet measured

Public repository, Buzz Developer Preview, method fixed 2026-08-19

Can Buzz agents open PRs?

Block states the chat-to-PR path as the product's core loop on buzz.xyz, verified 2026-08-19. Depth at preview stage is what this run measures: how much of the path is one continuous surface and how much is a hand-off back to your terminal.

What is the method?

One issue, one repository, one operator, timed hops.

  • Repository: a public repository we control, so the PR is real and reversible
  • Issue: a bug with a reproducible failure, not a cosmetic change
  • Hops timed: discussion to plan, plan to first code, code to review, review to PR open
  • Recorded: every point the operator leaves Buzz for another tool

What would move the score?

Developer workflow fit is Buzz's highest cell at 8.0 with a MED confidence flag, because the loop is documented and not yet exercised on our bench. A completed run either raises the flag to HIGH at the same value or moves the value. Both outcomes publish.

Related

Method fixed 2026-08-19 · DESK REVIEW · PROVISIONAL