From 2d68849ca2dbeef432d5d928efff368c167b6e66 Mon Sep 17 00:00:00 2001 From: Pranav Raja Date: Mon, 17 Aug 2026 11:18:39 -0700 Subject: [PATCH] docs(ci): name the two QA preview suites in the /benchmark dispatcher header MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The suite argument is validated dynamically against `suites/*.toml` on nearai/benchmarks main, so nothing here gates a new suite — but that also means nothing here mentions one either, and `qa-automation-preview` is the suite most worth reaching for when changing reborn's integration or automation behaviour. Test PR for the /benchmark path on those suites; not for merge unless useful. Co-Authored-By: Claude Opus 5 (1M context) --- .github/workflows/nearai-bench.yml | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/.github/workflows/nearai-bench.yml b/.github/workflows/nearai-bench.yml index dabd8a2a07..91d828cf6e 100644 --- a/.github/workflows/nearai-bench.yml +++ b/.github/workflows/nearai-bench.yml @@ -10,6 +10,12 @@ # Setup / usage docs: # https://github.com/nearai/benchmarks/blob/main/docs/ci/README.md # +# The suite name is validated against `suites/*.toml` on nearai/benchmarks main, so a new +# suite is dispatchable as soon as it lands there — no allowlist to update here. Two worth +# knowing about for reborn's integration/automation behaviour: +# /benchmark qa-automation-preview 10 routine tasks; the axis that separates harnesses +# /benchmark qa-preview the full 38-task QA corpus +# # nearai/benchmarks is a PRIVATE repo (since 2026-07-22, PR #291). A # cross-repo reusable-workflow `uses:` call (the previous approach here) is # structurally incapable of working for a public caller -> private callee: