ci(free-tests): run the free suite on ubicloud-standard-4 (week-1 revert rule for #3617) (#3710)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

.github/workflows/free-tests.yml

@@ -37,7 +37,7 @@ on:
 # push runs never see workflow_dispatch inputs, so free-plan resolves these
 # (or a dispatch override) into outputs that free-suite reads.
 env:
-  FREE_SUITE_RUNNER: ubicloud-standard-2
+  FREE_SUITE_RUNNER: ubicloud-standard-4
   FREE_SUITE_SHARDS: '20'
 
 # Keyed on the PR number, not head_ref: a bare branch name carries no fork

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## [1.92.1.0] - 2026-10-31
+
+**The free suite moves from `ubicloud-standard-2` to `ubicloud-standard-4`, still at 20 shards, because standard-2 made Free Tests slower than its revert rule allows.** #3617's week-1 check measured three paired rounds at main 5ba425af. On standard-2 the slowest shard took 380, 396 and 388 s, against 314, 327 and 336 s on standard-8, and runs finished in 514 to 526 s against 405 to 434 s. On standard-4 the slowest shard took 323, 335 and 317 s, inside standard-8's 313 to 336 s range from the same day's paired runs. Per shard, standard-4 runs 10 to 15 % slower than standard-8 and costs about 57 % as much per run, so it keeps most of the saving without slowing the run's slowest shard.
+
+### Changed
+- **`.github/workflows/free-tests.yml`**: `FREE_SUITE_RUNNER` is `ubicloud-standard-4`. The shard count, the dispatch overrides and the `RESOURCES:` line are unchanged.
+- **docs/TESTING_INTERNALS.md** names the new runner.
+
 ## [1.92.0.0] - 2026-10-31
 
 **Merge trains work for any repository, not just gstack.** Until now a train needed gstack's own eval workflow, gated only its tip and needed a bisect to find a culprit, so peer crews like gbrain and Truther merged one PR at a time and folded main-red fixes together by hand (GSTA-157).

VERSION

@@ -1 +1 @@
-1.92.0.0
+1.92.1.0

agents-digest/gstack-AGENTS.md

@@ -1,4 +1,4 @@
-# gstack digest v1.92.0.0 — regenerate/re-copy after upgrading gstack
+# gstack digest v1.92.1.0 — regenerate/re-copy after upgrading gstack
 
 Behavioral rules from gstack (https://github.com/garrytan/gstack), compressed
 for agent hosts without a full skill install. The full skills add workflows,

docs/TESTING_INTERNALS.md

@@ -199,7 +199,7 @@ one duration-packed `--ci-plan`, 20 ordinary `--ci-run` shards plus a separate
 exclusive-fixture shard, and a `--ci-verify` aggregate. CI shards still run on
 independent machines without ordering unrelated jobs. Two workflow-level
 constants in `free-tests.yml` set that size for every event: `FREE_SUITE_RUNNER`
-(`ubicloud-standard-2`) and `FREE_SUITE_SHARDS` (`'20'`), which `free-plan`
+(`ubicloud-standard-4`) and `FREE_SUITE_SHARDS` (`'20'`), which `free-plan`
 resolves into the shards' runner and shard count. A `workflow_dispatch` run may
 override them with `free_runner` (`ubicloud-standard-2`, `-4` or `-8`) and
 `free_shards` (1–40) for runner-sizing experiments only; its job summary says
@@ -1540,7 +1540,7 @@ refresh to a maintainer.
   `scripts/ubicloud/setup-free-suite.sh` and the Windows seed through
   `windows-free-tests.yml`'s `record_durations` path. The recorder stays on
   `ubicloud-standard-8` while the free-suite shards run on `FREE_SUITE_RUNNER`
-  (`ubicloud-standard-2`); both are Ubicloud standard runners. When either recording adds
+  (`ubicloud-standard-4`); both are Ubicloud standard runners. When either recording adds
   or drops a test file, it force-updates the `bot/free-duration-seeds` branch,
   opens or refreshes one seed PR whose body states the ratchet result, and
   dispatches Free Tests, Windows Free Tests and Quality gate on it. Those

package.json

@@ -1,6 +1,6 @@
 {
   "name": "gstack",
-  "version": "1.92.0",
+  "version": "1.92.1",
   "description": "Garry's Stack — Claude Code skills + fast headless browser. One repo, one install, entire AI engineering workflow.",
   "license": "MIT",
   "type": "module",