feat: support live replay test reporters

[thymikee]

Summary

Enable custom replay test reporters to receive live progress events while agent-device test is running, with stream-aware stdout/stderr reporter context.

Update the default reporter to render through the same live reporter path and document a custom emoji reporter example.

Touched 16 files; scope stayed within replay test reporting, daemon client progress forwarding, CLI command wiring, tests, and docs.

Validation

pnpm exec vitest run src/__tests__/cli-network.test.ts src/utils/__tests__/daemon-client.test.ts passed: 2 files, 46 tests.

pnpm typecheck passed.

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.
2026-06-30 19:10 UTC

[github-actions]

Size Report

Metric	Base	Current	Diff
JS raw	1.4 MB	1.4 MB	+2.4 kB
JS gzip	450.2 kB	450.4 kB	+186 B
npm tarball	549.1 kB	549.4 kB	+305 B
npm unpacked	1.9 MB	1.9 MB	+2.4 kB

Startup median (7 runs, lower is better):

Scenario	Base	Current	Diff
CLI --version	28.3 ms	27.6 ms	-0.6 ms
CLI --help	47.4 ms	46.9 ms	-0.6 ms

Top changed chunks:

Chunk	Raw diff	Gzip diff
dist/src/cli.js	+16.1 kB	+5.1 kB
dist/src/find.js	+8.3 kB	+3.3 kB
dist/src/495.js	-5.7 kB	-2.1 kB
dist/src/2948.js	+9 B	+82 B

[thymikee]

Reviewed #959 against the live replay test reporter contract and the command path from agent-device test through CLI transport wrapping, daemon progress streaming, socket/HTTP client readers, reporter runtime dispatch, custom reporter loading, public exports, tests, and docs.

The progress path now keeps daemon progress parsing transport-level and routes replay-test events into the same reporter instances that render final output. Generic command progress still falls back to stderr, JSON mode still avoids implicit default reporter output, and the documented reporter exit-code behavior now prevents a custom reporter from masking a failing suite with exit 0. No linked issue is attached, so I reviewed against the PR summary and existing replay reporter docs.

Checks are green, including Typecheck, Unit Tests, Integration Tests, Smoke Tests, Fallow Code Quality, Lint & Format, Swift Runner Unit Compile, Web Platform Smoke, and deploy-preview. I do not see an actionable blocker; this is ready for maintainer review.

[thymikee]

The head changed after my previous review (e6e9ed9 -> 4669dfd) and this run still has pending checks, so I removed ready-for-human for now. I will re-review the current head once checks settle and then restore the label if there are still no blockers.

[thymikee]

Re-reviewed the current clean #959 head (42981b3) after the force-push and pending-check window.

The live reporter path still matches the intended contract: daemon progress parsing remains transport-level, replay-test events route into the same reporter runtime used for final output, custom reporters see semantic hooks (onSuiteStart, onTestStart, onTestStep, onTestResult), JSON mode still avoids implicit reporter output, and generic command progress still falls back to stderr. The docs now describe the stream/context API without adding unsupported public helper exports.

Checks are green, including Typecheck, Unit Tests, Integration Tests, Smoke Tests, Coverage, Fallow Code Quality, Lint & Format, Swift Runner Unit Compile, Web Platform Smoke, iOS Runner Swift Compatibility, and deploy-preview. I do not see an actionable blocker; ready for maintainer review.

[thymikee]

Head changed after the prior readiness review (now at 8e41006), and CI is still in progress. I removed ready-for-human for now; please wait for checks to settle and re-review this head before restoring the label.

[thymikee]

Review finding on the current clean head aa4d7bf:

ReplayTestReporter live hooks are typed as void | Promise<void> (src/replay/test/reporters/types.ts), but runReplayTestReporterHook starts returned promises and only attaches .catch(...) without waiting for them. renderReplayTestResponse then immediately runs onSuiteEnd, so a stateful async reporter can receive final output before its live onTestResult/onTestStep work has completed. That makes the public reporter contract racy even though the type advertises async hooks.

Please either make live hooks explicitly synchronous in the public type/docs, or track/flush pending live-hook promises before onSuiteEnd and add a regression test with an async live hook whose state is consumed by onSuiteEnd. Checks are green, but I would not restore ready-for-human until this contract is resolved.