diff --git a/crates/perry-runtime/Cargo.toml b/crates/perry-runtime/Cargo.toml
index 3d3f52b7ce..b7878e8136 100644
--- a/crates/perry-runtime/Cargo.toml
+++ b/crates/perry-runtime/Cargo.toml
@@ -15,7 +15,15 @@ crate-type = ["rlib", "staticlib"]
 # actually needs (see optimized_libs.rs), so the heavy subsystems below
 # (regex engine, Temporal, URL/IDNA, normalize, segmenter) are *opt-in per
 # app* and absent from binaries that never use them.
-default = ["full", "regex-engine", "temporal", "url-engine", "string-normalize", "intl-segmenter"]
+default = ["full", "regex-engine", "temporal", "url-engine", "string-normalize", "intl-segmenter", "diagnostics"]
+# Cold-path diagnostic JSON serializers (~67 KB of code + the `serde_json`
+# pulled only by them, which dead-strips when unreferenced): GC cycle telemetry
+# (`PERRY_GC_DIAG`), typed-feedback trace dump (`PERRY_TYPED_FEEDBACK`), the v8
+# heap-snapshot builder (`v8.getHeapSnapshot`/`writeHeapSnapshot`), and
+# `process.report`. None are on a hot path. The env-driven dev diagnostics
+# degrade gracefully when off (auto-optimize leaves this off unless the program
+# uses a heap-snapshot / `process.report` API, which the compiler detects).
+diagnostics = []
 # The user's regular-expression engine (`regex` + `fancy-regex`, ~1.2 MB of
 # DFA/NFA machinery). A program that never evaluates a regex literal, `RegExp`,
 # a regex-coercing string method, or a glob API can't produce a RegExp at
diff --git a/crates/perry-runtime/src/arena/mod.rs b/crates/perry-runtime/src/arena/mod.rs
index c8b32b9b0f..777a1fb155 100644
--- a/crates/perry-runtime/src/arena/mod.rs
+++ b/crates/perry-runtime/src/arena/mod.rs
@@ -53,6 +53,8 @@ pub use allocators::{
 pub(crate) use allocators::{arena_alloc_gc_old_excluding_pages, arena_alloc_gc_survivor};
 
 // walk.rs
+#[cfg(feature = "diagnostics")]
+pub(crate) use walk::ArenaRegionTelemetry;
 pub use walk::{
     arena_block_count, arena_in_use_bytes, arena_total_bytes, arena_walk_objects,
     arena_walk_objects_addr_sorted, arena_walk_objects_filtered,
@@ -61,8 +63,8 @@ pub use walk::{
 };
 pub(crate) use walk::{
     arena_block_snapshots, arena_telemetry_snapshot, general_block_in_recent_window,
-    ArenaBlockSnapshot, ArenaObjectCursor, ArenaObjectCursorBuilder, ArenaRegionTelemetry,
-    ArenaTelemetrySnapshot, ArenaWalkOrder,
+    ArenaBlockSnapshot, ArenaObjectCursor, ArenaObjectCursorBuilder, ArenaTelemetrySnapshot,
+    ArenaWalkOrder,
 };
 
 // reset.rs
diff --git a/crates/perry-runtime/src/arena/walk.rs b/crates/perry-runtime/src/arena/walk.rs
index 3942d09183..23c7b43ed4 100644
--- a/crates/perry-runtime/src/arena/walk.rs
+++ b/crates/perry-runtime/src/arena/walk.rs
@@ -376,6 +376,7 @@ pub(crate) struct ArenaRegionTelemetry {
 }
 
 #[derive(Clone, Copy, Default)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(crate) struct ArenaTelemetrySnapshot {
     pub(crate) arena: ArenaRegionTelemetry,
     pub(crate) survivor0: ArenaRegionTelemetry,
diff --git a/crates/perry-runtime/src/gc/cycle.rs b/crates/perry-runtime/src/gc/cycle.rs
index 6119c19ac4..8e48cdfd13 100644
--- a/crates/perry-runtime/src/gc/cycle.rs
+++ b/crates/perry-runtime/src/gc/cycle.rs
@@ -14,6 +14,7 @@ pub(super) enum GcCyclePhase {
 }
 
 impl GcCyclePhase {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) const fn as_str(self) -> &'static str {
         match self {
diff --git a/crates/perry-runtime/src/gc/malloc.rs b/crates/perry-runtime/src/gc/malloc.rs
index 9d5427badd..ea84ae386f 100644
--- a/crates/perry-runtime/src/gc/malloc.rs
+++ b/crates/perry-runtime/src/gc/malloc.rs
@@ -30,6 +30,7 @@ impl MallocKindTelemetry {
         }
     }
 
+    #[cfg(feature = "diagnostics")]
     pub(super) fn reset_cycle_deltas(&mut self) {
         self.allocated_count = 0;
         self.allocated_bytes = 0;
@@ -273,6 +274,7 @@ impl MallocState {
             counters.copied_minor_validation_lookups.saturating_add(1);
     }
 
+    #[cfg(feature = "diagnostics")]
     pub(super) fn take_kind_telemetry(
         &mut self,
     ) -> [MallocKindTelemetry; MALLOC_KIND_BUCKET_COUNT] {
diff --git a/crates/perry-runtime/src/gc/mod.rs b/crates/perry-runtime/src/gc/mod.rs
index d5516a92a5..ab6a35fa88 100644
--- a/crates/perry-runtime/src/gc/mod.rs
+++ b/crates/perry-runtime/src/gc/mod.rs
@@ -58,7 +58,9 @@ mod cycle;
 use cycle::*;
 mod verify;
 pub use verify::*;
+#[cfg(feature = "diagnostics")]
 mod heap_snapshot;
+#[cfg(feature = "diagnostics")]
 pub use heap_snapshot::gc_build_v8_heap_snapshot_json;
 
 pub fn gc_collect_minor() -> u64 {
diff --git a/crates/perry-runtime/src/gc/oldgen.rs b/crates/perry-runtime/src/gc/oldgen.rs
index 64f0dadc99..6f05d67272 100644
--- a/crates/perry-runtime/src/gc/oldgen.rs
+++ b/crates/perry-runtime/src/gc/oldgen.rs
@@ -107,6 +107,7 @@ impl Default for EvacuationPolicyDecision {
 }
 
 #[derive(Clone, Copy, Default)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct SweepTraceStats {
     pub(super) dead_bytes: u64,
     // Compatibility alias for dead_bytes.
diff --git a/crates/perry-runtime/src/gc/policy.rs b/crates/perry-runtime/src/gc/policy.rs
index 687283e3bd..b83dbd84f0 100644
--- a/crates/perry-runtime/src/gc/policy.rs
+++ b/crates/perry-runtime/src/gc/policy.rs
@@ -365,6 +365,7 @@ pub(super) enum GcCollectionKind {
 }
 
 impl GcCollectionKind {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) fn as_str(self) -> &'static str {
         match self {
@@ -394,6 +395,7 @@ pub(super) enum GcTriggerKind {
 }
 
 impl GcTriggerKind {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) fn as_str(self) -> &'static str {
         match self {
@@ -466,6 +468,7 @@ impl DeferredGcRequest {
 }
 
 #[derive(Clone, Copy)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct GcStepSnapshot {
     pub(super) arena_step_bytes: usize,
     pub(super) next_arena_trigger_bytes: usize,
diff --git a/crates/perry-runtime/src/gc/roots.rs b/crates/perry-runtime/src/gc/roots.rs
index b05d0d743b..696b428fd5 100644
--- a/crates/perry-runtime/src/gc/roots.rs
+++ b/crates/perry-runtime/src/gc/roots.rs
@@ -143,6 +143,7 @@ pub(super) enum ConservativeStackScanDecision {
 }
 
 impl ConservativeStackScanDecision {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) const fn as_str(self) -> &'static str {
         match self {
diff --git a/crates/perry-runtime/src/gc/telemetry.rs b/crates/perry-runtime/src/gc/telemetry.rs
index f6941cd0ee..91b4499836 100644
--- a/crates/perry-runtime/src/gc/telemetry.rs
+++ b/crates/perry-runtime/src/gc/telemetry.rs
@@ -15,6 +15,7 @@ thread_local! {
 }
 
 #[derive(Clone, Copy, Default)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct RememberedSetTraceStats {
     pub(super) entries_scanned: usize,
     pub(super) valid_roots: usize,
@@ -98,6 +99,7 @@ pub(super) enum CopiedMinorFallbackReason {
 }
 
 impl CopiedMinorFallbackReason {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) const fn as_str(self) -> &'static str {
         match self {
@@ -218,6 +220,7 @@ impl RootSourceSlotTraceStats {
 }
 
 #[derive(Clone, Copy, Default)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct NativeStackFallbackTraceStats {
     pub(super) decision: ConservativeStackScanDecision,
     pub(super) scanned: bool,
@@ -524,6 +527,7 @@ pub(super) enum AllocatorMaintenanceStatus {
 }
 
 impl AllocatorMaintenanceStatus {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) const fn as_str(self) -> &'static str {
         match self {
@@ -543,6 +547,7 @@ pub(super) enum AllocatorMaintenanceReason {
 }
 
 impl AllocatorMaintenanceReason {
+    #[cfg(feature = "diagnostics")]
     #[inline]
     pub(super) const fn as_str(self) -> &'static str {
         match self {
@@ -554,6 +559,7 @@ impl AllocatorMaintenanceReason {
 }
 
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct AllocatorMaintenanceEvent {
     pub(super) status: AllocatorMaintenanceStatus,
     pub(super) reason: AllocatorMaintenanceReason,
@@ -561,10 +567,12 @@ pub(super) struct AllocatorMaintenanceEvent {
 }
 
 #[derive(Clone, Copy, Debug, Default, Eq, PartialEq)]
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct AllocatorMaintenanceTrace {
     pub(super) malloc_trim: Option<AllocatorMaintenanceEvent>,
 }
 
+#[cfg_attr(not(feature = "diagnostics"), allow(dead_code))]
 pub(super) struct GcCycleTrace {
     pub(super) collection_kind: GcCollectionKind,
     pub(super) trigger_kind: GcTriggerKind,
@@ -719,6 +727,7 @@ impl GcCycleTrace {
         }
     }
 
+    #[cfg(feature = "diagnostics")]
     pub(super) fn into_json(mut self, steps_after: GcStepSnapshot) -> serde_json::Value {
         self.capture_layout_scans();
         self.debt.record(GcDebtSnapshot::current());
@@ -964,6 +973,7 @@ impl GcCycleTrace {
         })
     }
 
+    #[cfg(feature = "diagnostics")]
     pub(super) fn emit(self, steps_after: GcStepSnapshot) {
         let event = self.into_json(steps_after);
         #[cfg(test)]
@@ -972,8 +982,16 @@ impl GcCycleTrace {
             eprintln!("{line}");
         }
     }
+
+    #[cfg(not(feature = "diagnostics"))]
+    pub(super) fn emit(self, _steps_after: GcStepSnapshot) {
+        eprintln!(
+            "[gc] cycle (diagnostics feature disabled — rebuild without --no-default-features for JSON trace)"
+        );
+    }
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn debt_snapshot_json(snapshot: GcDebtSnapshot) -> serde_json::Value {
     serde_json::json!({
         "arena_debt_bytes": snapshot.arena_debt_bytes,
@@ -982,6 +1000,7 @@ pub(super) fn debt_snapshot_json(snapshot: GcDebtSnapshot) -> serde_json::Value
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn pause_budget_json(
     progress_kind: GcProgressKind,
     progress_budget: GcPauseBudget,
@@ -999,6 +1018,7 @@ pub(super) fn pause_budget_json(
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn pause_step_json(step: GcPauseStepTrace) -> serde_json::Value {
     let progress_budget = gc_progress_contract().budget_for(step.progress_kind);
     let within_soft_pause_target = progress_budget
@@ -1026,6 +1046,7 @@ pub(super) fn pause_step_json(step: GcPauseStepTrace) -> serde_json::Value {
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn allocator_maintenance_json(
     trace: AllocatorMaintenanceTrace,
     progress_kind: GcProgressKind,
@@ -1045,6 +1066,7 @@ pub(super) fn allocator_maintenance_json(
     })
 }
 
+#[cfg(feature = "diagnostics")]
 fn default_malloc_trim_maintenance(progress_kind: GcProgressKind) -> AllocatorMaintenanceEvent {
     if progress_kind.is_budgeted() {
         return AllocatorMaintenanceEvent {
@@ -1134,6 +1156,7 @@ pub(super) fn malloc_object_count() -> usize {
     MALLOC_STATE.with(|s| s.borrow().objects.len())
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn malloc_kind_telemetry_row(
     obj_type: u8,
     counters: MallocKindTelemetry,
@@ -1154,6 +1177,7 @@ pub(super) fn malloc_kind_telemetry_row(
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn root_source_slot_json(stats: RootSourceSlotTraceStats) -> serde_json::Value {
     serde_json::json!({
         "registered_scanners": stats.registered_scanners,
@@ -1164,6 +1188,7 @@ pub(super) fn root_source_slot_json(stats: RootSourceSlotTraceStats) -> serde_js
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn root_sources_json(stats: RootSourcesTraceStats) -> serde_json::Value {
     serde_json::json!({
         "compiled_shadow": root_source_slot_json(stats.compiled_shadow),
@@ -1183,6 +1208,7 @@ pub(super) fn root_sources_json(stats: RootSourcesTraceStats) -> serde_json::Val
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn malloc_kind_telemetry_json_from_snapshot(
     snapshot: [MallocKindTelemetry; MALLOC_KIND_BUCKET_COUNT],
 ) -> serde_json::Value {
@@ -1201,11 +1227,13 @@ pub(super) fn malloc_kind_telemetry_json_from_snapshot(
     serde_json::Value::Array(rows)
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn take_malloc_kind_telemetry_json() -> serde_json::Value {
     let snapshot = MALLOC_STATE.with(|s| s.borrow_mut().take_kind_telemetry());
     malloc_kind_telemetry_json_from_snapshot(snapshot)
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn arena_region_json(region: crate::arena::ArenaRegionTelemetry) -> serde_json::Value {
     serde_json::json!({
         "in_use_bytes": region.in_use_bytes,
@@ -1214,6 +1242,7 @@ pub(super) fn arena_region_json(region: crate::arena::ArenaRegionTelemetry) -> s
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn arena_snapshot_json(
     snapshot: crate::arena::ArenaTelemetrySnapshot,
 ) -> serde_json::Value {
@@ -1229,6 +1258,7 @@ pub(super) fn arena_snapshot_json(
     })
 }
 
+#[cfg(feature = "diagnostics")]
 pub(super) fn steps_json(before: GcStepSnapshot, after: GcStepSnapshot) -> serde_json::Value {
     serde_json::json!({
         "arena_step_bytes": {
diff --git a/crates/perry-runtime/src/gc/types.rs b/crates/perry-runtime/src/gc/types.rs
index ced071c3e0..724a233779 100644
--- a/crates/perry-runtime/src/gc/types.rs
+++ b/crates/perry-runtime/src/gc/types.rs
@@ -623,6 +623,7 @@ pub(crate) unsafe fn gc_type_finalize_unmarked_payload(obj_type: u8, user_ptr: *
     }
 }
 
+#[cfg(feature = "diagnostics")]
 #[inline]
 pub(super) fn gc_type_name(obj_type: u8) -> &'static str {
     gc_type_info(obj_type).map_or("unknown", |info| info.name)
diff --git a/crates/perry-runtime/src/node_v8.rs b/crates/perry-runtime/src/node_v8.rs
index 3d1de9e375..b6bdd53065 100644
--- a/crates/perry-runtime/src/node_v8.rs
+++ b/crates/perry-runtime/src/node_v8.rs
@@ -334,7 +334,12 @@ pub extern "C" fn js_v8_cached_data_version_tag() -> f64 {
 #[no_mangle]
 pub extern "C" fn js_v8_get_heap_snapshot(options: f64) -> f64 {
     validate_heap_snapshot_options(options);
+    #[cfg(feature = "diagnostics")]
     let json = crate::gc::gc_build_v8_heap_snapshot_json();
+    // OFF stub: the compiler enables `diagnostics` whenever a program uses the
+    // v8 heap-snapshot APIs, so this branch is unreachable in practice.
+    #[cfg(not(feature = "diagnostics"))]
+    let json = String::from("{}");
     snapshot_readable_stream(&json)
 }
 
@@ -352,7 +357,12 @@ pub extern "C" fn js_v8_write_heap_snapshot(filename: f64, options: f64) -> f64
         }
     };
     validate_heap_snapshot_options(options);
+    #[cfg(feature = "diagnostics")]
     let json = crate::gc::gc_build_v8_heap_snapshot_json();
+    // OFF stub: unreachable in practice (compiler enables `diagnostics` when a
+    // program uses the v8 heap-snapshot APIs).
+    #[cfg(not(feature = "diagnostics"))]
+    let json = String::from("{}");
     match std::fs::write(&path, json.as_bytes()) {
         Ok(()) => string_value(&path),
         Err(err) => unsafe {
diff --git a/crates/perry-runtime/src/process.rs b/crates/perry-runtime/src/process.rs
index c741b2cea2..d4e881f582 100644
--- a/crates/perry-runtime/src/process.rs
+++ b/crates/perry-runtime/src/process.rs
@@ -801,7 +801,12 @@ extern "C" fn process_report_function_write_report(
     let filename = module_value_to_string(file_arg)
         .filter(|s| !s.is_empty())
         .unwrap_or_else(process_report_default_filename);
+    // OFF stub: unreachable in practice (the compiler enables `diagnostics`
+    // whenever a program references `process.report`).
+    #[cfg(feature = "diagnostics")]
     let report_json = process_report_json_string("API", Some(&filename));
+    #[cfg(not(feature = "diagnostics"))]
+    let report_json = String::from("{}");
     if let Err(err) = std::fs::write(&filename, report_json) {
         crate::fs::validate::throw_type_error_with_code(
             &format!("Failed to write diagnostic report to {filename}: {err}"),
@@ -1282,6 +1287,7 @@ fn node_platform_name() -> &'static str {
     }
 }
 
+#[cfg(feature = "diagnostics")]
 fn process_report_json_string(trigger: &str, filename: Option<&str>) -> String {
     let args: Vec<String> = std::env::args().collect();
     let command_line = if args.is_empty() {
diff --git a/crates/perry-runtime/src/typed_feedback.rs b/crates/perry-runtime/src/typed_feedback.rs
index 6b3b2ecae2..88385a1264 100644
--- a/crates/perry-runtime/src/typed_feedback.rs
+++ b/crates/perry-runtime/src/typed_feedback.rs
@@ -5,10 +5,11 @@
 //! has actually seen at runtime.
 
 use std::collections::{BTreeMap, HashMap};
-use std::sync::{
-    atomic::{AtomicBool, Ordering},
-    LazyLock, Mutex,
-};
+#[cfg(any(feature = "diagnostics", test))]
+use std::sync::atomic::AtomicBool;
+#[cfg(any(feature = "diagnostics", test))]
+use std::sync::atomic::Ordering;
+use std::sync::{LazyLock, Mutex};
 
 use crate::array::ArrayHeader;
 use crate::object::ObjectHeader;
@@ -21,6 +22,7 @@ const POLYMORPHIC_CAP: usize = 4;
 
 static REGISTRY: LazyLock<Mutex<TypedFeedbackRegistry>> =
     LazyLock::new(|| Mutex::new(TypedFeedbackRegistry::default()));
+#[cfg(any(feature = "diagnostics", test))]
 static TRACE_DUMPED: AtomicBool = AtomicBool::new(false);
 
 #[cfg(not(test))]
@@ -961,9 +963,9 @@ pub use guards::{
 
 #[path = "typed_feedback/trace.rs"]
 mod trace;
-pub use trace::{
-    js_typed_feedback_maybe_dump_trace, typed_feedback_snapshot, typed_feedback_trace_json,
-};
+pub use trace::typed_feedback_snapshot;
+#[cfg(feature = "diagnostics")]
+pub use trace::{js_typed_feedback_maybe_dump_trace, typed_feedback_trace_json};
 
 fn hash_bytes(bytes: &[u8]) -> u64 {
     let mut h = 0xcbf2_9ce4_8422_2325u64;
diff --git a/crates/perry-runtime/src/typed_feedback/trace.rs b/crates/perry-runtime/src/typed_feedback/trace.rs
index 3ae58a68dd..c67acbfb4f 100644
--- a/crates/perry-runtime/src/typed_feedback/trace.rs
+++ b/crates/perry-runtime/src/typed_feedback/trace.rs
@@ -1,3 +1,4 @@
+#[cfg(feature = "diagnostics")]
 use std::path::{Path, PathBuf};
 
 use super::*;
@@ -225,6 +226,7 @@ pub fn typed_feedback_snapshot() -> TypedFeedbackSnapshot {
     snapshot
 }
 
+#[cfg(feature = "diagnostics")]
 pub fn typed_feedback_trace_json() -> serde_json::Value {
     let snapshot = typed_feedback_snapshot();
     serde_json::json!({
@@ -283,6 +285,7 @@ pub fn typed_feedback_trace_json() -> serde_json::Value {
     })
 }
 
+#[cfg(feature = "diagnostics")]
 fn typed_feedback_trace_path_from_env() -> Option<PathBuf> {
     let value = std::env::var("PERRY_TYPED_FEEDBACK_TRACE").ok()?;
     if value.is_empty() || value == "0" {
@@ -295,6 +298,7 @@ fn typed_feedback_trace_path_from_env() -> Option<PathBuf> {
     }
 }
 
+#[cfg(feature = "diagnostics")]
 fn ensure_parent_dir(path: &Path) -> std::io::Result<()> {
     if let Some(parent) = path.parent() {
         if !parent.as_os_str().is_empty() {
@@ -304,28 +308,38 @@ fn ensure_parent_dir(path: &Path) -> std::io::Result<()> {
     Ok(())
 }
 
+// The extern symbol must ALWAYS be compiled: codegen emits an unconditional
+// call to it from `main` (the program-exit trace-dump hook), so gating the
+// whole function breaks the final link under auto-optimize (`Undefined symbols:
+// _js_typed_feedback_maybe_dump_trace`). Only the JSON-building body is gated
+// behind `diagnostics`; with the feature off this is a no-op (the
+// `PERRY_TYPED_FEEDBACK` trace is a dev diagnostic, absent from size-optimized
+// binaries).
 #[no_mangle]
 pub extern "C" fn js_typed_feedback_maybe_dump_trace() {
-    let Some(path) = typed_feedback_trace_path_from_env() else {
-        return;
-    };
-    if TRACE_DUMPED.swap(true, Ordering::AcqRel) {
-        return;
-    }
-
-    let json = typed_feedback_trace_json();
-    let bytes = match serde_json::to_vec_pretty(&json) {
-        Ok(bytes) => bytes,
-        Err(err) => {
-            eprintln!("perry typed-feedback trace: failed to encode JSON: {err}");
+    #[cfg(feature = "diagnostics")]
+    {
+        let Some(path) = typed_feedback_trace_path_from_env() else {
+            return;
+        };
+        if TRACE_DUMPED.swap(true, Ordering::AcqRel) {
             return;
         }
-    };
-    if let Err(err) = ensure_parent_dir(&path).and_then(|_| std::fs::write(&path, bytes)) {
-        eprintln!(
-            "perry typed-feedback trace: failed to write {}: {err}",
-            path.display()
-        );
+
+        let json = typed_feedback_trace_json();
+        let bytes = match serde_json::to_vec_pretty(&json) {
+            Ok(bytes) => bytes,
+            Err(err) => {
+                eprintln!("perry typed-feedback trace: failed to encode JSON: {err}");
+                return;
+            }
+        };
+        if let Err(err) = ensure_parent_dir(&path).and_then(|_| std::fs::write(&path, bytes)) {
+            eprintln!(
+                "perry typed-feedback trace: failed to write {}: {err}",
+                path.display()
+            );
+        }
     }
 }
 
@@ -373,5 +387,6 @@ mod keep_typed_feedback {
     #[used] static K20: extern "C" fn(u64, i64, f64, f64) = js_typed_feedback_object_set_index_polymorphic;
     #[used] static K21: extern "C" fn(u64, *mut ObjectHeader, u32, *const crate::StringHeader, f64) = js_typed_feedback_object_set_unboxed_f64_field;
     #[used] static K22: extern "C" fn(u64, f64) -> f64 = js_typed_feedback_observe_helper_return;
+    #[cfg(feature = "diagnostics")]
     #[used] static K23: extern "C" fn() = js_typed_feedback_maybe_dump_trace;
 }
diff --git a/crates/perry/src/commands/compile/collect_modules.rs b/crates/perry/src/commands/compile/collect_modules.rs
index b3dbc254de..8ad98fb8a1 100644
--- a/crates/perry/src/commands/compile/collect_modules.rs
+++ b/crates/perry/src/commands/compile/collect_modules.rs
@@ -1927,6 +1927,23 @@ fn collect_module_finish(
         }
     }
 
+    // Detect heap-snapshot / `process.report` usage, the only user-facing APIs
+    // behind the `diagnostics` feature (~95 KB of cold-path JSON serializers +
+    // the `serde_json` pulled only by them). `v8.getHeapSnapshot` /
+    // `v8.writeHeapSnapshot` lower to `NativeMethodCall { method: "…" }`;
+    // `process.report.*` surfaces as `property: "report"`. The env-driven dev
+    // diagnostics (GC-diag / typed-feedback JSON) ride the same feature and
+    // degrade gracefully when off, so they need no detection.
+    {
+        let hir_debug: String = format!("{:?}{:?}", &hir_module.init, &hir_module.functions);
+        if hir_debug.contains("method: \"getHeapSnapshot\"")
+            || hir_debug.contains("method: \"writeHeapSnapshot\"")
+            || hir_debug.contains("property: \"report\"")
+        {
+            ctx.uses_diagnostics = true;
+        }
+    }
+
     // Detect readline usage via process.stdin raw/lifecycle methods. These
     // don't go through an `import 'readline'` statement, so the import-based
     // needs_stdlib detection above misses them.
diff --git a/crates/perry/src/commands/compile/link/mod.rs b/crates/perry/src/commands/compile/link/mod.rs
index b4d1279c6b..34ccb0e1ba 100644
--- a/crates/perry/src/commands/compile/link/mod.rs
+++ b/crates/perry/src/commands/compile/link/mod.rs
@@ -882,7 +882,11 @@ pub(super) fn build_and_run_link(
             "{}/toolchains/llvm/prebuilt/{}/bin/aarch64-linux-android24-clang{}",
             ndk_home,
             host_tag,
-            if cfg!(target_os = "windows") { ".cmd" } else { "" }
+            if cfg!(target_os = "windows") {
+                ".cmd"
+            } else {
+                ""
+            }
         );
         let stub_ok = Command::new(&ndk_clang)
             .args(["-c", "-fPIC", "-target", "aarch64-linux-android24"])
diff --git a/crates/perry/src/commands/compile/link/platform_cmd.rs b/crates/perry/src/commands/compile/link/platform_cmd.rs
index 05ca762936..5d86946e01 100644
--- a/crates/perry/src/commands/compile/link/platform_cmd.rs
+++ b/crates/perry/src/commands/compile/link/platform_cmd.rs
@@ -576,7 +576,11 @@ pub fn select_linker_command(
             "{}/toolchains/llvm/prebuilt/{}/bin/aarch64-linux-android24-clang{}",
             ndk_home,
             host_tag,
-            if cfg!(target_os = "windows") { ".cmd" } else { "" }
+            if cfg!(target_os = "windows") {
+                ".cmd"
+            } else {
+                ""
+            }
         );
         if !PathBuf::from(&clang).exists() {
             return Err(anyhow!("Android NDK clang not found at: {}", clang));
diff --git a/crates/perry/src/commands/compile/optimized_libs.rs b/crates/perry/src/commands/compile/optimized_libs.rs
index 256e1cf47b..83d58ba849 100644
--- a/crates/perry/src/commands/compile/optimized_libs.rs
+++ b/crates/perry/src/commands/compile/optimized_libs.rs
@@ -670,7 +670,7 @@ pub(super) fn build_optimized_libs(
     // Cheap djb2 — no need for the SipHash overhead.
     let target_str = target.unwrap_or("host");
     let key_input = format!(
-        "{}|{}|{}|wasm={}|regex={}|temporal={}|url={}|norm={}|seg={}|v={}",
+        "{}|{}|{}|wasm={}|regex={}|temporal={}|url={}|norm={}|seg={}|diag={}|v={}",
         feature_arg,
         panic_abort_safe,
         target_str,
@@ -680,6 +680,7 @@ pub(super) fn build_optimized_libs(
         ctx.uses_url,
         ctx.uses_string_normalize,
         ctx.uses_intl_segmenter,
+        ctx.uses_diagnostics,
         env!("CARGO_PKG_VERSION"),
     );
     let mut hash: u64 = 5381;
@@ -796,6 +797,13 @@ pub(super) fn build_optimized_libs(
     if ctx.uses_intl_segmenter {
         cross_features.push("perry-runtime/intl-segmenter".to_string());
     }
+    // Cold-path diagnostic JSON serializers (~95 KB incl. the `serde_json`
+    // pulled only by them) — enabled only when the program uses a heap-snapshot
+    // API or `process.report`. The env-driven GC/typed-feedback dev trace JSON
+    // ride this feature and stay off in size-optimized binaries.
+    if ctx.uses_diagnostics {
+        cross_features.push("perry-runtime/diagnostics".to_string());
+    }
     if !cross_features.is_empty() {
         cargo_cmd.arg("--features").arg(cross_features.join(","));
     }
diff --git a/crates/perry/src/commands/compile/types.rs b/crates/perry/src/commands/compile/types.rs
index 19338354d9..782cb99e6b 100644
--- a/crates/perry/src/commands/compile/types.rs
+++ b/crates/perry/src/commands/compile/types.rs
@@ -570,6 +570,14 @@ pub struct CompilationContext {
     /// `perry-runtime/intl-segmenter` (`unicode-segmentation`, ~73 KB of UAX #29
     /// grapheme/word/sentence tables). Other `Intl.*` APIs don't need it.
     pub uses_intl_segmenter: bool,
+    /// Whether any TS module uses a heap-snapshot API (`v8.getHeapSnapshot` /
+    /// `v8.writeHeapSnapshot`) or `process.report`. Gates
+    /// `perry-runtime/diagnostics` (the cold-path JSON serializers + the
+    /// `serde_json` pulled only by them, ~95 KB). The env-driven dev
+    /// diagnostics (`PERRY_GC_DIAG` JSON trace, typed-feedback trace dump) ride
+    /// the same feature and degrade gracefully when it's off, so they're absent
+    /// from size-optimized binaries unless one of these APIs is also used.
+    pub uses_diagnostics: bool,
     /// Whether `perry/thread` is imported. When true, the runtime must
     /// keep `panic = "unwind"` so that worker-thread panics translate to
     /// promise rejections via `catch_unwind` in `perry-runtime/src/thread.rs`
@@ -818,6 +826,7 @@ impl CompilationContext {
             uses_url: false,
             uses_string_normalize: false,
             uses_intl_segmenter: false,
+            uses_diagnostics: false,
             needs_thread: false,
             cross_module_class_field_types: HashMap::new(),
             min_windows_version: "10".to_string(),