build(sdk): route npm test through scripts/test.sh (split from #165)

[kjgbot]

The sdk half of #165, without its CI revert. Do not merge yet — one open question below.

What this is

#165 bundled two unrelated changes. Its .github/workflows/cloud-runtime-artifact.yml hunk reverted #153, #154 and #159 together: it restored ops/cargo.sh for the kernel test step, and dropped the analyzer-skip env. That wrapper redirects RUSTUP_HOME to $HOME/.relayflows-toolchain/rustup, which on a GitHub runner is empty, so the rustup shim has nothing to choose and the step dies with rustup could not choose a version of cargo to run. The wrapper exists for a cloud sandbox, which a runner is not.

This branch carries only the sdk tooling. .github/ is byte-identical to main:

$ git diff origin/main --stat -- .github/ | wc -l
0

sdk/package.json differs from main by exactly one line (the test script), so nothing else on that file is reverted despite #165 being based on an older main.

scripts/test.sh runs the same chain the inline script did, in the same order, with set -eu supplying the fail-fast the && chain gave. CI is unaffected either way — the workflow runs the expanded chain minus test:prep, not npm test.

Open question before merge

prune-test-build.mjs deletes every .map and .d.ts under sdk/dist, recursively. sdk/package.json declares:

"types": "./dist/index.d.ts"

So running npm test leaves the package's declared type entry pointing at a file that was just deleted. Nothing in the repo breaks today — the only in-repo consumers (ops/probes/**) import .js from dist, and surface does not depend on the sdk at all — so this is latent, not live. But the next TypeScript consumer of @relayflows/sdk gets a confusing failure whose cause is "someone ran the tests".

I did not change the script, because that would alter the contributed change rather than split it. Three ways forward, your call:

1. Prune only .map and leave .d.ts alone — smallest change, keeps the declared types intact.
2. Keep the prune and drop the types entry from package.json, if dist types are not meant to be consumed.
3. Merge as-is and accept the latent trap.

I would take (1): the stated motivation for pruning is propagated-tree file count in a cloud sandbox, and ops/cargo.sh's own measurements put that squarely on kernel/target/debug (4900 files) rather than on sdk/dist, so deleting the type surface buys very little.

Supersedes the sdk half of #165. #165 itself should be closed once this lands — its remaining content is the CI revert, which we do not want.

🤖 Generated with Claude Code

https://claude.ai/code/session_01FtQSAcGDta5VH9xiZFT4sR

[coderabbitai]

Warning

Review limit reached

Next included review available in 51 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Free

Run ID: 6f29e4e6-49ef-411c-af91-36858459daa8

📥 Commits

Reviewing files that changed from the base of the PR and between 733f496 and 3e7ee75.

📒 Files selected for processing (3)

• sdk/package.json
• sdk/scripts/prune-test-build.mjs
• sdk/scripts/test.sh

---

Note

🎁 Summarized by CodeRabbit Free

Your organization is on the Free plan. CodeRabbit will generate a high-level summary and a walkthrough for each pull request. For a comprehensive line-by-line review, please upgrade your subscription to CodeRabbit Essentials by visiting https://app.coderabbit.ai/settings/billing.

_{Comment @coderabbitai help to get the list of available commands.}