fix(agents): disable agy implementation dispatch

Codex handoff could not write .agents in its sandbox, so the orchestrator applied the intended fail-fast change: agy remains available for dry-run prompt inspection but non-dry-run dispatch exits before worktree or PR side effects.

.agents/README.md

@@ -24,7 +24,7 @@ bin/dev          CLI: `dev implement <issue#> --harness <name> [--dry-run]`
 - **Groom** (interactive, Claude): `/groom <issue#>` runs a seeded grill-me → writes a
   Ready spec into the issue body → sets Effort/Priority → leaves Stage at `Research`.
   You review and flip to `Ready`.
-- **Implement** (any harness): `.agents/bin/dev implement <issue#> --harness <name>`
+- **Implement** (recommended harnesses: `codex` or `cursor`): `.agents/bin/dev implement <issue#> --harness <name>`
   (or `/implement …` in Claude to supervise) → isolated worktree → runs the harness
   headless on the spec → build gate → opens a PR → sets Stage `In Review`.
 - **Review + merge**: use the `review` skill from a harness different from the
@@ -42,8 +42,17 @@ local adapter state.
 ## Adding / fixing a harness
 
 Edit one function in `harnesses.sh`: `harness_<name> <workdir> <promptfile>`, running the
-tool non-interactively in `<workdir>` on the prompt. `claude`/`codex`/`opencode` are
+tool non-interactively in `<workdir>` on the prompt. For current implementation
-wired; `pi`/`cursor` are stubs — confirm their headless flags before trusting.
+dispatch, prefer `codex` or `cursor`; `claude` and `opencode` remain available, and
+`pi` still needs flag verification before trusting.
+
+`agy` / Antigravity is deliberately disabled for non-dry-run dispatch as of
+2026-07-04. Phase 1 testing found the headless path unreliable: with
+`--new-project` it ignored the prompt and tried to scaffold, while without it the CLI
+could resume a stale conversation and hang past the print timeout. Keep using
+`.agents/bin/dev implement <issue#> --harness agy --dry-run` for prompt inspection
+only; real dispatch should use `codex` or `cursor` until a fresh Antigravity
+headless invocation is verified and documented.
 
 Always `--dry-run` a new harness first: it renders the exact prompt and plan, touching
 nothing (no worktree, PR, or state change).

.agents/bin/dev

@@ -1,7 +1,7 @@
 #!/usr/bin/env bash
 # .agents/bin/dev — box-box agentic dev CLI. Works from any harness's shell.
 #
-#   dev implement <issue#> --harness <claude|codex|opencode|pi|cursor> [--dry-run] [--base <branch>]
+#   dev implement <issue#> --harness <codex|cursor|claude|opencode|pi|agy> [--dry-run] [--base <branch>]
 #
 # Grooming is driven interactively via the Claude Code /groom skill; this CLI covers
 # the implement lane (dispatch a Ready issue to a harness → worktree → gate → PR).
@@ -15,7 +15,7 @@ box-box dev CLI
 
   dev implement <issue#> --harness <name> [--dry-run] [--base <branch>]
 
-harnesses: claude, codex, opencode (supported) · pi, cursor (verify flags in .agents/harnesses.sh)
+harnesses: codex, cursor (recommended) · claude, opencode (available) · agy (dry-run only, disabled for dispatch) · pi (verify flags)
   --dry-run   render the prompt + plan, touch nothing (no worktree/PR/state change)
   --base      base branch for the worktree/PR (default: main)
 EOF

.agents/harnesses.sh

@@ -11,6 +11,18 @@
 
 # ---- MUST-HAVE ----
 
+harness_disabled_reason() {              # <name> -> reason on stdout; 0 means disabled
+  case "$1" in
+    agy)
+      cat <<'EOF'
+agy is disabled for implementation dispatch as of 2026-07-04: Antigravity headless mode was observed to ignore prompts with --new-project, resume stale conversations without it, and hang past print timeouts. Use codex or cursor until a fresh headless invocation is verified.
+EOF
+      return 0
+      ;;
+    *) return 1 ;;
+  esac
+}
+
 harness_claude() {                      # Claude Code — print mode, auto-accept edits
   local dir="$1" prompt="$2"
   ( cd "$dir" && claude -p "$(cat "$prompt")" --permission-mode acceptEdits )
@@ -31,14 +43,11 @@ harness_cursor() {                      # Cursor CLI agent — composer-2.5, hea
   ( cd "$dir" && cursor-agent -p "$(cat "$prompt")" --model composer-2.5 --force --trust )
 }
 
-harness_agy() {                         # Antigravity CLI — UNRELIABLE headless (2026-07): with
+harness_agy() {                         # Antigravity CLI — disabled until headless is verified
-  # --new-project it ignores the prompt and asks to scaffold a project; without it, it
+  local reason
-  # resumes the previous conversation (silently keeping its old model — --model only
+  reason="$(harness_disabled_reason agy)"
-  # applies to new conversations) and can hang past the print timeout. Do not trust for
+  echo "harness_agy: $reason" >&2
-  # dispatch until fixed upstream; verify with a trivial prompt first.
+  return 2
-  local dir="$1" prompt="$2"
-  ( cd "$dir" && agy --print --print-timeout 60m \
-      --model="Gemini 3.1 Pro (High)" --dangerously-skip-permissions "$(cat "$prompt")" )
 }
 
 # ---- NICE-TO-HAVE (verify the exact invocation for your version before trusting) ----

.agents/lib/dispatch.sh

@@ -52,6 +52,15 @@ dispatch() { # <issue#> <harness> [--dry-run] [--base <branch>]
   if ! declare -f "harness_$harness" >/dev/null 2>&1; then
     echo "no adapter for harness '$harness' — add harness_$harness() to .agents/harnesses.sh" >&2; return 2
   fi
+  if [ "$dry" != 1 ] && declare -f harness_disabled_reason >/dev/null 2>&1; then
+    local disabled_reason
+    if disabled_reason="$(harness_disabled_reason "$harness")"; then
+      echo "harness '$harness' is disabled for non-dry-run dispatch." >&2
+      echo "  $disabled_reason" >&2
+      echo "  Use --dry-run for prompt inspection, or dispatch with --harness codex/cursor." >&2
+      return 2
+    fi
+  fi
 
   local repo_root title body slug branch wt prompt
   repo_root="$(git rev-parse --show-toplevel)" || return 1

.agents/test/dispatch_disabled_harness_test.sh

@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
+cd "$ROOT"
+
+# shellcheck source=/dev/null
+source "$ROOT/.agents/lib/dispatch.sh"
+
+issue_title() { echo "Disable agy harness smoke"; }
+issue_body() { echo "## Spec"; echo; echo "Smoke prompt body"; }
+get_field() { echo "Ready"; }
+set_stage() { echo "unexpected set_stage $*" >&2; return 99; }
+run_gate() { echo "unexpected run_gate $*" >&2; return 99; }
+
+unexpected_git_file="$(mktemp "${TMPDIR:-/tmp}/boxbox-agy-git.XXXX")"
+rm -f "$unexpected_git_file"
+git() {
+  if [ "${1:-}" = "rev-parse" ]; then
+    command git "$@"
+    return
+  fi
+  echo "unexpected git $*" >&2
+  touch "$unexpected_git_file"
+  return 99
+}
+
+set +e
+non_dry_output="$(dispatch 47 agy 2>&1)"
+non_dry_status=$?
+set -e
+
+[ "$non_dry_status" -eq 2 ] || {
+  echo "expected agy non-dry-run to exit 2, got $non_dry_status" >&2
+  echo "$non_dry_output" >&2
+  exit 1
+}
+[[ "$non_dry_output" == *"harness 'agy' is disabled"* ]] || {
+  echo "expected disabled-harness message" >&2
+  echo "$non_dry_output" >&2
+  exit 1
+}
+[ ! -e "$unexpected_git_file" ] || {
+  echo "agy non-dry-run reached git before failing" >&2
+  echo "$non_dry_output" >&2
+  exit 1
+}
+
+dry_output="$(dispatch 47 agy --dry-run 2>&1)"
+[[ "$dry_output" == *"[dry-run] no worktree / harness / PR / state change"* ]] || {
+  echo "expected agy dry-run to render dispatch preview" >&2
+  echo "$dry_output" >&2
+  exit 1
+}
+[[ "$dry_output" == *"Smoke prompt body"* ]] || {
+  echo "expected agy dry-run prompt body" >&2
+  echo "$dry_output" >&2
+  exit 1
+}

cmd/main.go

@@ -27,7 +27,6 @@ func main() {
 	ingestMeeting := flag.Int("ingest-meeting", 0, "Ingest meeting metadata and Race Hub datasets for all sessions")
 	ingestSession := flag.Int("ingest-session", 0, "Ingest Race Hub datasets for a session key")
 	ingestNews := flag.Bool("ingest-news", false, "Refresh RSS/Atom paddock briefing feeds")
-	prefetchTrackOutlines := flag.Int("prefetch-track-outlines", 0, "Warm the web track-outline cache for a season year (for web-only hosts, run before --web so /api/v1/track-outline can serve live maps)")
 	dryRun := flag.Bool("dry-run", false, "Preview ingestion without writing domain rows")
 	force := flag.Bool("force", false, "Re-ingest datasets even if already tracked in the session_coverage table as completed")
 	coverageYear := flag.Int("coverage", 0, "Show season coverage report for the given year")
@@ -77,21 +76,11 @@ func main() {
 	if *ingestNews {
 		ingestFlags++
 	}
-	if *prefetchTrackOutlines != 0 {
-		ingestFlags++
-	}
 	if ingestFlags > 0 {
 		if ingestFlags > 1 {
-			fmt.Fprintln(os.Stderr, "box-box: only one of --ingest-year, --backfill-season, --ingest-meeting, --ingest-session, --ingest-news, or --prefetch-track-outlines may be set")
+			fmt.Fprintln(os.Stderr, "box-box: only one of --ingest-year, --backfill-season, --ingest-meeting, --ingest-session, or --ingest-news may be set")
 			os.Exit(1)
 		}
-		if *prefetchTrackOutlines != 0 {
-			if err := runTrackOutlinePrefetch(client, *prefetchTrackOutlines); err != nil {
-				fmt.Fprintf(os.Stderr, "box-box track outline prefetch error: %v\n", err)
-				os.Exit(1)
-			}
-			return
-		}
 		if *ingestNews {
 			if err := runNewsIngestion(*dryRun, *dbPath); err != nil {
 				fmt.Fprintf(os.Stderr, "box-box ingest error: %v\n", err)
@@ -185,35 +174,6 @@ func runIngestion(client *api.OpenF1Client, year, meetingKey, sessionKey int, fo
 	return err
 }
 
-func runTrackOutlinePrefetch(client *api.OpenF1Client, year int) error {
-	log.SetOutput(os.Stderr)
-
-	fmt.Fprintf(os.Stderr, "track outlines: warming HTTP cache %s for %d\n", api.DefaultCacheDBPath(), year)
-
-	meetings, err := client.GetMeetingsForYear(year)
-	if err != nil {
-		return fmt.Errorf("fetch meetings for %d: %w", year, err)
-	}
-
-	result := client.PrefetchTrackOutlinesForYear(year, meetings)
-	fmt.Printf(
-		"track outlines %d: cached %d/%d unique circuit(s) before, %d/%d after; %d skipped, %d fetched, %d failed\n",
-		result.Year,
-		result.CachedBefore,
-		result.UniqueCircuits,
-		result.CachedAfter,
-		result.UniqueCircuits,
-		result.Skipped,
-		result.Fetched,
-		result.Failed,
-	)
-
-	if result.CachedAfter == 0 {
-		return fmt.Errorf("cached zero track outlines for %d", year)
-	}
-	return nil
-}
-
 func runNewsIngestion(dryRun bool, dbPath string) error {
 	log.SetOutput(os.Stderr)
 

internal/api/cache.go

@@ -94,12 +94,6 @@ func cacheDBPath() string {
 	return filepath.Join(".cache", "box-box", "cache.db")
 }
 
-// DefaultCacheDBPath returns the HTTP cache database path used by the OpenF1
-// client in both TUI and web modes.
-func DefaultCacheDBPath() string {
-	return cacheDBPath()
-}
-
 // ttlForURL determines the appropriate TTL based on the URL pattern.
 // Returns 0 (CacheTTLForever) for historical data that will never change.
 func ttlForURL(url string) time.Duration {

internal/api/openf1.go

@@ -623,19 +623,6 @@ func (c *OpenF1Client) GetTeamRadio(sessionKey, driverNumber int) ([]models.Team
 // to maximise the chance of finding data quickly.
 var candidateDrivers = []int{1, 11, 44, 16, 55, 4, 14, 63, 81, 24}
 
-// TrackOutlinePrefetchResult summarizes a season track-outline cache warming
-// run. Counts are scoped to the unique non-zero circuit keys in the provided
-// meeting list.
-type TrackOutlinePrefetchResult struct {
-	Year           int
-	UniqueCircuits int
-	CachedBefore   int
-	CachedAfter    int
-	Skipped        int
-	Fetched        int
-	Failed         int
-}
-
 // PrefetchTrackOutlines fetches GPS location data for every circuit in the
 // provided meeting list and stores it in the cache so the track map tab can
 // render during live sessions when the free-tier API is locked.
@@ -645,58 +632,28 @@ type TrackOutlinePrefetchResult struct {
 // Errors per-circuit are silently ignored — this is a best-effort operation
 // and must never block or crash the main UI.
 func (c *OpenF1Client) PrefetchTrackOutlines(meetings []models.Meeting) {
-	year := time.Now().Year()
-	for _, m := range meetings {
-		if m.Year != 0 {
-			year = m.Year
-			break
-		}
-	}
-	_ = c.PrefetchTrackOutlinesForYear(year, meetings)
-}
-
-// PrefetchTrackOutlinesForYear fetches and caches track outlines for unique
-// circuits in the provided meeting list, storing them under the explicit season
-// year. Unlike PrefetchTrackOutlines, it returns accounting suitable for CLI
-// cache-warming workflows.
-func (c *OpenF1Client) PrefetchTrackOutlinesForYear(year int, meetings []models.Meeting) TrackOutlinePrefetchResult {
 	const maxWorkers = 3
 
-	result := TrackOutlinePrefetchResult{Year: year}
+	year := time.Now().Year()
-	uniqueByCircuit := make(map[int]models.Meeting)
+
-	var unique []models.Meeting
+	// Filter to meetings that need fetching.
+	var pending []models.Meeting
 	for _, m := range meetings {
 		if m.CircuitKey == 0 {
 			continue
 		}
-		if _, exists := uniqueByCircuit[m.CircuitKey]; exists {
-			continue
-		}
-		uniqueByCircuit[m.CircuitKey] = m
-		unique = append(unique, m)
-	}
-
-	result.UniqueCircuits = len(unique)
-
-	// Filter to meetings that need fetching.
-	var pending []models.Meeting
-	for _, m := range unique {
 		if _, ok := c.cache.GetTrackOutline(m.CircuitKey, year); ok {
-			result.CachedBefore++
-			result.Skipped++
 			continue // already cached for this season
 		}
 		pending = append(pending, m)
 	}
 
 	if len(pending) == 0 {
-		result.CachedAfter = result.CachedBefore
+		return
-		return result
 	}
 
 	sem := make(chan struct{}, maxWorkers)
 	var wg sync.WaitGroup
-	var mu sync.Mutex
 
 	for _, mtg := range pending {
 		mtg := mtg // capture
@@ -705,34 +662,19 @@ func (c *OpenF1Client) PrefetchTrackOutlinesForYear(year int, meetings []models.
 		go func() {
 			defer wg.Done()
 			defer func() { <-sem }()
-			ok := c.prefetchCircuit(mtg, year)
+			c.prefetchCircuit(mtg, year)
-			mu.Lock()
-			if ok {
-				result.Fetched++
-			} else {
-				result.Failed++
-			}
-			mu.Unlock()
 		}()
 	}
 
 	wg.Wait()
-
-	for _, m := range unique {
-		if _, ok := c.cache.GetTrackOutline(m.CircuitKey, year); ok {
-			result.CachedAfter++
-		}
-	}
-
-	return result
 }
 
 // prefetchCircuit fetches the track outline for a single meeting and stores it.
 // It prefers completed sessions (past date_end) so the data is full and stable.
-func (c *OpenF1Client) prefetchCircuit(mtg models.Meeting, year int) bool {
+func (c *OpenF1Client) prefetchCircuit(mtg models.Meeting, year int) {
 	sessions, err := c.GetSessionsForMeeting(int(mtg.MeetingKey))
 	if err != nil || len(sessions) == 0 {
-		return false
+		return
 	}
 
 	// Pick the best session: prefer a completed race, then any session with
@@ -754,7 +696,7 @@ func (c *OpenF1Client) prefetchCircuit(mtg models.Meeting, year int) bool {
 		}
 	}
 	if bestSession == nil {
-		return false
+		return
 	}
 
 	// Try candidate drivers in order until we find one with enough points.
@@ -764,7 +706,7 @@ func (c *OpenF1Client) prefetchCircuit(mtg models.Meeting, year int) bool {
 			continue
 		}
 		// Store under the circuit key for this year and stop.
-		return c.cache.SetTrackOutline(mtg.CircuitKey, year, locs) == nil
+		_ = c.cache.SetTrackOutline(mtg.CircuitKey, year, locs)
+		return
 	}
-	return false
 }

internal/api/track_outline_prefetch_test.go

@@ -1,109 +0,0 @@
-package api
-
-import (
-	"encoding/json"
-	"net/http"
-	"net/http/httptest"
-	"strconv"
-	"testing"
-	"time"
-
-	"github.com/AmanTahiliani/box-box/internal/models"
-)
-
-func newTrackOutlineTestClient(t *testing.T, srvURL string) *OpenF1Client {
-	t.Helper()
-	t.Setenv("HOME", t.TempDir())
-	t.Setenv("XDG_CACHE_HOME", t.TempDir())
-
-	c := NewOpenF1Client(srvURL, 5*time.Second)
-	c.pacer = &requestPacer{}
-	t.Cleanup(func() { _ = c.Close() })
-	return c
-}
-
-func TestPrefetchTrackOutlinesForYearSkipsCachedAndWritesLocations(t *testing.T) {
-	var sessionsByMeeting = map[string][]models.Session{
-		"202": {
-			{
-				SessionKey:  9002,
-				SessionName: "Race",
-				MeetingKey:  202,
-				CircuitKey:  2,
-				DateEnd:     "2026-01-01T12:00:00+00:00",
-			},
-		},
-	}
-	var sessionsRequested []string
-	var locationsRequested []string
-
-	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		switch r.URL.Path {
-		case "/v1/sessions":
-			meetingKey := r.URL.Query().Get("meeting_key")
-			sessionsRequested = append(sessionsRequested, meetingKey)
-			_ = json.NewEncoder(w).Encode(sessionsByMeeting[meetingKey])
-		case "/v1/location":
-			sessionKey := r.URL.Query().Get("session_key")
-			driverNumber := r.URL.Query().Get("driver_number")
-			locationsRequested = append(locationsRequested, sessionKey+"/"+driverNumber)
-			_ = json.NewEncoder(w).Encode(testLocations(9002, 1, 51))
-		default:
-			t.Fatalf("unexpected request path %s", r.URL.Path)
-		}
-	}))
-	defer srv.Close()
-
-	client := newTrackOutlineTestClient(t, srv.URL)
-	if err := client.Cache().SetTrackOutline(1, 2026, testLocations(9001, 1, 51)); err != nil {
-		t.Fatalf("SetTrackOutline() error = %v", err)
-	}
-
-	result := client.PrefetchTrackOutlinesForYear(2026, []models.Meeting{
-		{MeetingKey: 101, Year: 2026, Circuit: models.Circuit{CircuitKey: 1}},
-		{MeetingKey: 202, Year: 2026, Circuit: models.Circuit{CircuitKey: 2}},
-		{MeetingKey: 303, Year: 2026, Circuit: models.Circuit{CircuitKey: 2}},
-	})
-
-	if result.UniqueCircuits != 2 {
-		t.Fatalf("UniqueCircuits = %d, want 2", result.UniqueCircuits)
-	}
-	if result.CachedBefore != 1 || result.Skipped != 1 || result.Fetched != 1 || result.Failed != 0 || result.CachedAfter != 2 {
-		t.Fatalf("unexpected result: %+v", result)
-	}
-	if got, want := len(sessionsRequested), 1; got != want {
-		t.Fatalf("sessions requested %d time(s), want %d: %v", got, want, sessionsRequested)
-	}
-	if sessionsRequested[0] != "202" {
-		t.Fatalf("requested meeting %s, want 202", sessionsRequested[0])
-	}
-	if got, want := len(locationsRequested), 1; got != want {
-		t.Fatalf("locations requested %d time(s), want %d: %v", got, want, locationsRequested)
-	}
-	if locationsRequested[0] != "9002/1" {
-		t.Fatalf("requested location %s, want 9002/1", locationsRequested[0])
-	}
-
-	locs, ok := client.Cache().GetTrackOutline(2, 2026)
-	if !ok {
-		t.Fatal("expected circuit 2 outline to be cached")
-	}
-	if len(locs) != 51 {
-		t.Fatalf("cached %d locations, want 51", len(locs))
-	}
-}
-
-func testLocations(sessionKey, driverNumber, count int) []models.Location {
-	locs := make([]models.Location, count)
-	for i := range locs {
-		locs[i] = models.Location{
-			Date:         "2026-01-01T12:00:" + strconv.Itoa(i%60) + "+00:00",
-			DriverNumber: driverNumber,
-			MeetingKey:   202,
-			SessionKey:   sessionKey,
-			X:            float64(i),
-			Y:            float64(i * 2),
-		}
-	}
-	return locs
-}