feat: expose fetch option on ManagementClient

[tsushanth]

Why

Closes #1330.

The internal fetcher layer added `BaseClientOptions.fetch?: typeof fetch` in #1238 and routes it through `fetcherImpl(args.fetchFn ?? …)`. The public wrapper, however, both omitted `fetch` from `ManagementClientOptions` and explicitly deleted it from the incoming options bag, so library consumers had no supported way to provide their own HTTP transport — even though the SDK is already capable of using one. Typical use cases the reporter and downstream Auth0 users have called out: proxies / corporate egress, retry middlewares, OpenTelemetry instrumentation, and Cloudflare Workers / Bun targets where the global `fetch` needs swapping.

What

• Drop `"fetch"` from the `Omit<FernClient.Options, …>` list so the field is part of the public `ManagementClientOptions` surface.
• Stop deleting `_options.fetch` in the `ManagementClient` constructor. The internal `fetcher` is still stripped (it's a Fern implementation detail), with a comment explaining why and linking back to this issue.

Tests

src/management/tests/unit/management-client-fetch.test.ts - two cases:

• Constructs a ManagementClient with a custom fetch, fires client.rules.list(), and asserts the custom function was invoked against tenant.auth0.com/api/v2/rules and that the parsed return value matches the payload it served.
• Constructs a ManagementClient without fetch to confirm the default global-fetch route is unchanged.

Compat

Source-compatible: existing call sites that didn't pass `fetch` see no behavioural difference. Callers can now opt in:

```ts
new ManagementClient({
domain: 'tenant.auth0.com',
token: '...',
fetch: myInstrumentedFetch,
});
```

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.75%. Comparing base (a823058) to head (ad61319).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1352      +/-   ##
==========================================
- Coverage   89.76%   89.75%   -0.01%     
==========================================
  Files         431      431              
  Lines       20108    20107       -1     
  Branches     9782     9782              
==========================================
- Hits        18049    18048       -1     
  Misses       2059     2059              

Flag	Coverage Δ
alltests	89.75% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/management/wrapper/ManagementClient.ts	100.00% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[harshithRai]

Hi @tsushanth,

Thanks for tracking this down and putting together a clean fix 🙌!
The root cause analysis in the PR description is spot on, and the source change in ManagementClient.ts is exactly right.

A few things to fix before we merge:

Chores on the GitHub side

1. Unverified commits: The repo requires signed commits. Please rebase with GPG/SSH-signed commits (git rebase --exec 'git commit --amend --no-edit -S' HEAD~N) before we proceed.
2. Branch is out of date: please rebase on master (ideally as part of the signed-commit rebase above, so it's one step).

Code feedback

3. Test file location:
   The new test lands in tests/management/, but wrapper-level tests live in src/management/tests/unit/ (see management-client-custom-domain.test.ts). These two suites have different environments; tests/ runs ESM against the compiled package, src/management/tests/unit/ runs CJS directly against source. The wrapper test belongs in the unit suite.

4. JSDoc on fetch:
   fetch inherits a doc comment from BaseClientOptions in BaseClient.ts, but that's Fern-generated and thinner than the proxy/retry/instrumentation/Workers language from the issue; worth adding an explicit JSDoc override on ManagementClientOptions.

The core fix is solid, just these before it's ready to merge 🙂.

[harshithRai]

Hi @tsushanth! Thanks again for the original fix and root-cause work here. Since the review comments were still open and the branch had fallen behind the v6 release, I've pushed directly to this branch (edits-by-maintainers was enabled) to get it over the line.

What I added on top of your commit:

• Rebased onto current master (was behind the v6 release).
• Moved the wrapper test into the unit suite (src/management/tests/unit/) alongside the other wrapper tests, and made it assert on the real rules.list() return value.
• Added an explicit enriched JSDoc on the public fetch field (proxies, retries, instrumentation, Workers/Bun).
• Fixed the adjacent class-level example that still used the fetcher: callback (which the constructor deletes); now uses fetch:.

Your core fix is unchanged and remains on your commit.

Your contribution is much appreciated 🙌

[harshithRai]

After a discussion with @frederikprijck on approach: for consistency with the other Auth0 SDKs, we're going to expose a wrapped fetcher (Auth0FetcherSupplier) rather than a raw fetch. The other SDKs omit both fetch and fetcher from the Fern options and re-expose a stable fetcher wrapper, hiding the codegen internals (example).

Since that's a different design and a larger change, it'll land as a separate PR rather than reworking this branch. Closing this one in favor of that - bu thanks @tsushanth for the root-cause work on #1330 and the original fix; it's what surfaced the whole issue and pointed at exactly the right layer.

Credit carried over to the new PR.