🐛 fix: preserve PR attribution before Hive identity footer

changelog.d/fixed-10686-attribution-identity-footer.md

@@ -0,0 +1 @@
+- Most effective models now recognizes contributor PR attribution followed by Hive's identity footer, restoring merged counts and ranking eligibility on rescan while still rejecting attribution followed by ordinary prose ([#10686](https://github.com/hivecommons/hive/issues/10686)).

src/pkg/dashboard/contribute_effective_models_test.go

@@ -2,6 +2,7 @@ package dashboard
 
 import (
 	"encoding/json"
+	"fmt"
 	"net/http"
 	"net/http/httptest"
 	"os"
@@ -53,6 +54,42 @@ func TestAggregateContributeEffectiveModelsThresholdRankingAndPrivacy(t *testing
 	}
 }
 
+func TestAggregateContributeEffectiveModelsIdentityFooterAttribution(t *testing.T) {
+	now := time.Date(2026, 10, 5, 12, 0, 0, 0, time.UTC)
+	body := "Done\n\n— hive: backend=codex model=gpt-6-astra effort=medium\n\n---\n🐝 **Hive Agent**: `contributor` | **SHA:** `671737a3b`"
+	var prs []ghpkg.PullRequest
+	var runs []TaskRunRecord
+	for number := 1; number <= effectiveModelsDefaultMinMergedPRs; number++ {
+		meta, ok := ghpkg.ParseAttributionTrailer(body)
+		if !ok {
+			t.Fatal("production contributor PR body lost its attribution")
+		}
+		prs = append(prs, ghpkg.PullRequest{
+			Repo: "org/repo", Number: number, HiveAttributed: ok,
+			HiveBackend: meta.Backend, HiveModel: meta.Model,
+			CreatedAt: now, MergedAt: now, Rework: ghpkg.PRReworkStats{FirstPass: true},
+		})
+		runs = append(runs, TaskRunRecord{
+			TS: now.Format(time.RFC3339), Username: "contributor", Backend: "codex", Model: "gpt-6-astra",
+			Outcome: outcomeCompleted, PRVerified: true, PRURL: fmt.Sprintf("https://github.com/org/repo/pull/%d", number),
+		})
+	}
+	for _, filter := range []string{"all", "contributor"} {
+		below := aggregateContributeEffectiveModels(prs[:4], runs, "30d", filter, effectiveModelsDefaultMinMergedPRs, now)
+		if len(below.Ranked) != 0 || len(below.Insufficient) != 1 || below.Insufficient[0].MergedPRs != 4 {
+			t.Fatalf("%s below threshold = %+v", filter, below)
+		}
+		got := aggregateContributeEffectiveModels(prs, runs, "30d", filter, effectiveModelsDefaultMinMergedPRs, now)
+		if len(got.Ranked) != 1 || len(got.Insufficient) != 0 {
+			t.Fatalf("%s at threshold = %+v", filter, got)
+		}
+		row := got.Ranked[0]
+		if row.Model != "gpt-6-astra" || row.Backend != "codex" || row.MergedPRs != 5 || row.PRs != 5 || row.EffectivenessRank != 1 {
+			t.Fatalf("%s attribution/ranking = %+v", filter, row)
+		}
+	}
+}
+
 func TestHandleContributeEffectiveModelsPublicAggregate(t *testing.T) {
 	s := covApiServer(t)
 	now := time.Now()

src/pkg/github/attribution.go

@@ -296,11 +296,18 @@ func HasAttributionTrailer(body string) bool {
 	return strings.Contains(body, AttributionTrailerPrefix)
 }
 
+// attributionIdentityFooter matches only the terminal block emitted by
+// bin/gh-wrapper.sh's _identity_footer, not arbitrary trailing Markdown.
+var attributionIdentityFooter = regexp.MustCompile("\\n[ \\t]*---[ \\t]*\\n🐝 \\*\\*Hive Agent\\*\\*:(?: `[^`\\n]+`)?(?: \\| \\*\\*Instance:\\*\\* `[^`\\n]+`)? \\| \\*\\*SHA:\\*\\* `[^`\\n]+`[ \\t\\n]*$")
+
 // ParseAttributionTrailer extracts the launch metadata from the final visible
-// `— hive:` line in a PR body. Mentions of the template earlier in the body are
-// deliberately ignored so docs and quoted examples do not count as agent PRs.
+// `— hive:` line in a PR body, optionally followed by Hive's identity footer.
+// Mentions of the template earlier in the body are deliberately ignored so
+// docs and quoted examples do not count as agent PRs.
 func ParseAttributionTrailer(body string) (InvocationMeta, bool) {
-	lines := strings.Split(strings.ReplaceAll(body, "\r\n", "\n"), "\n")
+	body = strings.ReplaceAll(body, "\r\n", "\n")
+	body = attributionIdentityFooter.ReplaceAllString(body, "")
+	lines := strings.Split(body, "\n")
 	trailer := ""
 	for i := len(lines) - 1; i >= 0; i-- {
 		line := strings.TrimSpace(lines[i])

src/pkg/github/attribution_identity_footer_test.go

@@ -0,0 +1,80 @@
+package github
+
+import (
+	"context"
+	"io"
+	"log/slog"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	gh "github.com/google/go-github/v72/github"
+)
+
+const contributorAttributionBody = "Done\n\n— hive: backend=codex model=gpt-6-astra effort=medium\n\n---\n🐝 **Hive Agent**: `contributor` | **SHA:** `671737a3b`"
+
+func TestParseAttributionTrailerWithIdentityFooter(t *testing.T) {
+	for _, body := range []string{
+		contributorAttributionBody,
+		contributorAttributionBody + "\n\n",
+		strings.ReplaceAll(contributorAttributionBody, "\n", "\r\n"),
+		strings.ReplaceAll(contributorAttributionBody, " | **SHA:**", " | **Instance:** `hive-bold-fox` | **SHA:**"),
+		strings.ReplaceAll(contributorAttributionBody, " `contributor`", ""),
+	} {
+		meta, ok := ParseAttributionTrailer(body)
+		if !ok || meta.Backend != "codex" || meta.Model != "gpt-6-astra" {
+			t.Errorf("ParseAttributionTrailer(%q) = %+v, %v", body, meta, ok)
+		}
+	}
+}
+
+func TestParseAttributionTrailerDoesNotScanPastOrdinaryContent(t *testing.T) {
+	for _, body := range []string{
+		contributorAttributionBody + "\nMore prose",
+		strings.ReplaceAll(contributorAttributionBody, "\n\n---", "\nMore prose\n\n---"),
+		strings.ReplaceAll(contributorAttributionBody, "\n\n---", "\n\n"),
+		strings.ReplaceAll(contributorAttributionBody, "**SHA:**", "**Other:**"),
+		strings.ReplaceAll(contributorAttributionBody, "— hive:", "> — hive:"),
+		"```text\n" + contributorAttributionBody + "\n```",
+		strings.ReplaceAll(contributorAttributionBody, "\n\n---", "\n```\n\n---"),
+		contributorAttributionBody + "\n\n---\n🐝 **Hive Agent**: `contributor` | **SHA:** `unknown`",
+	} {
+		if meta, ok := ParseAttributionTrailer(body); ok {
+			t.Errorf("accepted nonterminal or quoted attribution in %q: %+v", body, meta)
+		}
+	}
+}
+
+func TestCollectAttributedPRsWithIdentityFooter(t *testing.T) {
+	pr := &gh.PullRequest{Number: gh.Ptr(10480), Body: gh.Ptr(contributorAttributionBody)}
+	prs := collectAttributedPRsFromList("hivecommons/hive", []*gh.PullRequest{pr})
+	if len(prs) != 1 || !prs[0].HiveAttributed || prs[0].HiveBackend != "codex" || prs[0].HiveModel != "gpt-6-astra" {
+		t.Fatalf("attributed PRs = %+v", prs)
+	}
+}
+
+func TestFetchAttributedClosedPRsWithIdentityFooter(t *testing.T) {
+	now := time.Now().UTC().Truncate(time.Second)
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/repos/o/r/pulls" || r.URL.Query().Get("state") != "closed" {
+			t.Errorf("unexpected request: %s", r.URL)
+			http.NotFound(w, r)
+			return
+		}
+		pr := attributedClosedPR(10480, now)
+		pr["body"] = contributorAttributionBody
+		pr["merged_at"] = now.Format(time.RFC3339)
+		writeJSON(t, w, []map[string]any{pr})
+	}))
+	defer srv.Close()
+	c := NewClientForTest(srv.URL, "o", []string{"r"}, slog.New(slog.NewTextHandler(io.Discard, nil)))
+	prs, err := c.fetchAttributedClosedPRs(context.Background(), "o", "r", "r")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(prs) != 1 || prs[0].HiveBackend != "codex" || prs[0].HiveModel != "gpt-6-astra" || !prs[0].MergedAt.Equal(now) {
+		t.Fatalf("merged attributed PRs = %+v", prs)
+	}
+}