diff --git a/.changeset/retry-sigsegv.md b/.changeset/retry-sigsegv.md
new file mode 100644
index 00000000000..5a53c351efe
--- /dev/null
+++ b/.changeset/retry-sigsegv.md
@@ -0,0 +1,5 @@
+---
+"@trigger.dev/core": patch
+---
+
+Retry `TASK_PROCESS_SIGSEGV` task crashes under the user's retry policy instead of failing the run on the first segfault. SIGSEGV in Node tasks is frequently non-deterministic (native addon races, JIT/GC interaction, near-OOM in native code, host issues), so retrying on a fresh process often succeeds. The retry is gated by the task's existing `retry` config + `maxAttempts` — same path `TASK_PROCESS_SIGTERM` and uncaught exceptions already use — so tasks without a retry policy still fail fast.
diff --git a/packages/core/src/v3/errors.ts b/packages/core/src/v3/errors.ts
index a538ca9357b..80124e1d386 100644
--- a/packages/core/src/v3/errors.ts
+++ b/packages/core/src/v3/errors.ts
@@ -199,6 +199,9 @@ export function truncateStack(stack: string | undefined): string {
   ].join("\n");
 }
 
+/**
+ * Truncates error messages that exceed MAX_MESSAGE_LENGTH to prevent OOM.
+ */
 export function truncateMessage(message: string | undefined): string {
   if (!message) return "";
   return message.length > MAX_MESSAGE_LENGTH
@@ -206,6 +209,10 @@ export function truncateMessage(message: string | undefined): string {
     : message;
 }
 
+/**
+ * Parses an unknown error into a TaskRunError structure.
+ * Handles InternalError, built-in Error, strings, and custom error objects.
+ */
 export function parseError(error: unknown): TaskRunError {
   if (isInternalError(error)) {
     return {
@@ -301,6 +308,10 @@ export function createJsonErrorObject(error: TaskRunError): SerializedError {
 }
 
 // Removes null characters and truncates oversized fields to prevent OOM
+/**
+ * Sanitizes TaskRunError by removing null bytes and truncating long fields.
+ * Used to clean errors before storage or transmission.
+ */
 export function sanitizeError(error: TaskRunError): TaskRunError {
   switch (error.type) {
     case "BUILT_IN_ERROR": {
@@ -351,6 +362,11 @@ export function sanitizeError(error: TaskRunError): TaskRunError {
   }
 }
 
+/**
+ * Determines whether an error should trigger a retry attempt.
+ * Returns true for errors that are retriable under the user's retry policy.
+ * Non-retriable errors (like OOM, SIGKILL_TIMEOUT) will fail the run immediately.
+ */
 export function shouldRetryError(error: TaskRunError): boolean {
   switch (error.type) {
     case "INTERNAL_ERROR": {
@@ -361,7 +377,6 @@ export function shouldRetryError(error: TaskRunError): boolean {
         case "CONFIGURED_INCORRECTLY":
         case "TASK_ALREADY_RUNNING":
         case "TASK_PROCESS_SIGKILL_TIMEOUT":
-        case "TASK_PROCESS_SIGSEGV":
         case "TASK_PROCESS_OOM_KILLED":
         case "TASK_PROCESS_MAYBE_OOM_KILLED":
         case "TASK_RUN_CANCELLED":
@@ -398,6 +413,7 @@ export function shouldRetryError(error: TaskRunError): boolean {
         case "TASK_RUN_UNCAUGHT_EXCEPTION":
         case "TASK_PROCESS_EXITED_WITH_NON_ZERO_CODE":
         case "TASK_PROCESS_SIGTERM":
+        case "TASK_PROCESS_SIGSEGV":
           return true;
 
         default:
@@ -419,6 +435,10 @@ export function shouldRetryError(error: TaskRunError): boolean {
   }
 }
 
+/**
+ * Checks if retry settings should be looked up for this error type.
+ * Some errors (like SIGSEGV, SIGTERM, uncaught exceptions) respect user retry config.
+ */
 export function shouldLookupRetrySettings(error: TaskRunError): boolean {
   switch (error.type) {
     case "INTERNAL_ERROR": {
@@ -448,6 +468,10 @@ export function shouldLookupRetrySettings(error: TaskRunError): boolean {
   }
 }
 
+/**
+ * Corrects error stack traces by normalizing file paths and removing noise.
+ * Used to make stack traces more readable in logs and error UI.
+ */
 export function correctErrorStackTrace(
   stackTrace: string,
   projectDir?: string,
@@ -490,6 +514,10 @@ function correctStackTraceLine(line: string, projectDir?: string, isDev?: boolea
   return line.trim();
 }
 
+/**
+ * Groups Zod validation issues by task index for better error reporting.
+ * Used when parsing task metadata fails.
+ */
 export function groupTaskMetadataIssuesByTask(tasks: any, issues: z.ZodIssue[]) {
   return issues.reduce(
     (acc, issue) => {
@@ -773,6 +801,10 @@ const findSignalInMessage = (message?: string, truncateLength = 100) => {
   }
 };
 
+/**
+ * Enhances TaskRunError with additional context like signals, OOM detection.
+ * Used to enrich errors before displaying or logging.
+ */
 export function taskRunErrorEnhancer(error: TaskRunError): EnhanceError<TaskRunError> {
   switch (error.type) {
     case "BUILT_IN_ERROR": {
diff --git a/packages/core/test/errors.test.ts b/packages/core/test/errors.test.ts
index dee6509d3a2..9a94366d845 100644
--- a/packages/core/test/errors.test.ts
+++ b/packages/core/test/errors.test.ts
@@ -1,5 +1,13 @@
 import { describe, it, expect } from "vitest";
-import { truncateStack, truncateMessage, parseError, sanitizeError } from "../src/v3/errors.js";
+import {
+  truncateStack,
+  truncateMessage,
+  parseError,
+  sanitizeError,
+  shouldRetryError,
+  shouldLookupRetrySettings,
+} from "../src/v3/errors.js";
+import type { TaskRunError } from "../src/v3/schemas/common.js";
 
 // Helper: build a fake stack with N frames
 function buildStack(messageLines: string[], frameCount: number): string {
@@ -238,3 +246,29 @@ describe("truncateStack message line bounding", () => {
     expect(result).toContain("...[truncated]");
   });
 });
+
+describe("shouldRetryError + shouldLookupRetrySettings", () => {
+  const internal = (code: string): TaskRunError =>
+    ({ type: "INTERNAL_ERROR", code } as TaskRunError);
+
+  it("retries SIGSEGV (changed from non-retriable) and looks up retry settings", () => {
+    const err = internal("TASK_PROCESS_SIGSEGV");
+    expect(shouldRetryError(err)).toBe(true);
+    expect(shouldLookupRetrySettings(err)).toBe(true);
+  });
+
+  it("retries SIGTERM via the same path", () => {
+    const err = internal("TASK_PROCESS_SIGTERM");
+    expect(shouldRetryError(err)).toBe(true);
+    expect(shouldLookupRetrySettings(err)).toBe(true);
+  });
+
+  it("still does not retry SIGKILL timeout", () => {
+    expect(shouldRetryError(internal("TASK_PROCESS_SIGKILL_TIMEOUT"))).toBe(false);
+  });
+
+  it("still does not retry OOM kills (handled by the separate machine-bump path)", () => {
+    expect(shouldRetryError(internal("TASK_PROCESS_OOM_KILLED"))).toBe(false);
+    expect(shouldRetryError(internal("TASK_PROCESS_MAYBE_OOM_KILLED"))).toBe(false);
+  });
+});