ci(#5441): enable parallel godog concurrency (default 12)

Makefile

@@ -197,7 +197,7 @@ e2e-test:
 	go test -tags e2e -v -count=1 -timeout 30m ./e2e/admin/
 
 behaviour-test:
-	go test -tags behaviour -v -count=1 -timeout 45m ./e2e/behaviour/
+	go test -tags behaviour -race -v -count=1 -timeout 45m ./e2e/behaviour/
 
 # Functional agent evals — run agents against ephemeral GitHub repos and judge results.
 # Required env: EVAL_ORG (GitHub org for ephemeral repos), plus GCP creds for Vertex AI.

docs/guides/dev/behaviour-drivers.md

@@ -32,7 +32,7 @@ The suite in `e2e/behaviour/suite_test.go` (or an external runner) acquires a po
 
 ### Install driver (v1 per-repo)
 
-Uses `fullsend inference provision <org>/test-repo` then `fullsend github setup <org>/test-repo --vendor --direct --skip-app-setup --runtime dummy` with the repo-scoped WIF provider from provision (`E2E_GCP_PROJECT_ID`). Pool orgs must already have shared GitHub Apps, org-level mint enrollment, and per-repo mint enrollment for `test-repo` (one-time GCP admin step on the hosted mint project). Numbered `test-repo-01` … `test-repo-12` names are also enrolled for planned parallelization; the driver does not select them yet. The driver does not run `fullsend admin install` or `fullsend mint enroll`. See [e2e-testing.md](e2e-testing.md#behaviour-tests-and-per-repo-mint-enrollment).
+Uses `fullsend inference provision <org>/test-repo` then `fullsend github setup <org>/test-repo --vendor --direct --skip-app-setup --runtime dummy` with the repo-scoped WIF provider from provision (`E2E_GCP_PROJECT_ID`). Pool orgs must already have shared GitHub Apps, org-level mint enrollment, and per-repo mint enrollment for `test-repo` (one-time GCP admin step on the hosted mint project). Numbered `test-repo-01` … `test-repo-12` are enrolled and actively leased from `world.RepoPool` during parallel scenario execution (see `GODOG_CONCURRENCY` in [behaviour-testing.md](behaviour-testing.md#parallel-execution)). The driver does not run `fullsend admin install` or `fullsend mint enroll`. See [e2e-testing.md](e2e-testing.md#behaviour-tests-and-per-repo-mint-enrollment).
 
 Teardown removes shim workflows, stale branches, and open fullsend PRs on `test-repo` via `pkg/e2etest.TeardownPerRepoInstall`.
 

docs/guides/dev/behaviour-testing.md

@@ -79,6 +79,30 @@ Use tags only for **exceptions** when a backend cannot run a scenario yet: `@ski
 make behaviour-test
 ```
 
+### Parallel execution
+
+The suite runs scenarios in parallel by default (`GODOG_CONCURRENCY=12`,
+matching the repo pool size). Each scenario gets its own `World` clone and
+leases a unique `test-repo-NN` from the pool, so no cross-scenario state
+is shared. The `behaviour-test` Make target includes `-race` to catch
+data races under concurrent execution.
+
+To adjust concurrency:
+
+```bash
+# Run at default concurrency (12)
+make behaviour-test
+
+# Run with explicit concurrency
+GODOG_CONCURRENCY=4 make behaviour-test
+
+# Serial mode for debugging
+GODOG_CONCURRENCY=1 make behaviour-test
+```
+
+Serial mode (`GODOG_CONCURRENCY=1`) is useful when debugging a single
+scenario or when `-v` output from multiple scenarios would interleave.
+
 In CI, the test runner mints cross-org `e2e` installation tokens via OIDC (same as admin e2e) for GitHub API operations. Triage workflows on the pool org's `test-repo` mint same-org `triage` tokens from vendored reusable workflows; those require per-repo mint enrollment (`PER_REPO_WIF_REPOS`) on the hosted mint project. Pool `test-repo` repos are enrolled once by a GCP admin — not during CI install. The install driver provisions repo-scoped inference WIF via `fullsend inference provision` before `github setup`. See [e2e-testing.md](e2e-testing.md#behaviour-tests-and-per-repo-mint-enrollment).
 
 ### Lazy create+install (`RepoEnsurer`)

e2e/behaviour/suite_test.go

@@ -5,6 +5,7 @@ package behaviour_test
 import (
 	"context"
 	"os"
+	"strconv"
 	"testing"
 
 	"github.com/cucumber/godog"
@@ -19,12 +20,27 @@ import (
 	"github.com/fullsend-ai/fullsend/pkg/e2etest"
 )
 
+// poolSize is the number of enrolled test-repo-NN repos in the pool org.
+// GODOG_CONCURRENCY must not exceed this — extra workers would block in
+// pool.Acquire with no warning because the pool org only has test-repo-01
+// through test-repo-12 with per-repo mint enrollment.
+const poolSize = 12
+
 func TestBehaviourSuite(t *testing.T) {
 	if testing.Short() {
 		t.Skip("skipping behaviour tests in short mode")
 	}
-	if c := os.Getenv("GODOG_CONCURRENCY"); c != "" && c != "1" {
-		t.Fatalf("behaviour suite does not support GODOG_CONCURRENCY=%q: per-scenario World isolation is in place but drivers (SCM, CI, Install) are shared by reference and have not been validated under -race with concurrent scenarios yet; see #5441 for parallel support", c)
+
+	concurrency := poolSize
+	if c := os.Getenv("GODOG_CONCURRENCY"); c != "" {
+		n, err := strconv.Atoi(c)
+		if err != nil || n < 1 {
+			t.Fatalf("GODOG_CONCURRENCY must be a positive integer, got %q", c)
+		}
+		concurrency = n
+	}
+	if concurrency > poolSize {
+		t.Fatalf("GODOG_CONCURRENCY=%d exceeds repo pool size %d", concurrency, poolSize)
 	}
 
 	cfg := env.LoadRunnerConfig()
@@ -65,7 +81,7 @@ func TestBehaviourSuite(t *testing.T) {
 		}
 	})
 
-	pool, err := world.NewRepoPool(12)
+	pool, err := world.NewRepoPool(poolSize)
 	if err != nil {
 		t.Fatalf("creating repo pool: %v", err)
 	}
@@ -96,7 +112,7 @@ func TestBehaviourSuite(t *testing.T) {
 			Paths:       []string{"features"},
 			TestingT:    t,
 			Tags:        os.Getenv("GODOG_TAGS"),
-			Concurrency: 1,
+			Concurrency: concurrency,
 		},
 	}
 	if st := suiteRunner.Run(); st != 0 {