v1.3.1

What's changed

• Fail-fast early bail — maxFailures (default 10) stops the run once that many tests have failed in total, prints the results gathered so far, and exits non-zero. Set 0 to run every test.
• Durable partial results — tests now run in ordered chunks via runByIds, so a protocolTimeout or crash mid-run prints the results from completed chunks instead of discarding the whole run.
• chunkSize (default 10) — how many tests run per browser call; smaller values make the failure limit and timeouts more granular.

maxFailures is on by default, so a run now stops after 10 total failures — raise it or set 0 to restore run-everything behavior. Exit codes are unchanged.

See #12 for details.

---

Big thanks to @evaleen for the feedback that drove this release! 🙏

v1.3.0

What's Changed

Features

• AI-friendly run output (#10) — the verbose output (config dump, per-test tree, status chatter) is replaced by a compact relay-style summary printed as the last output: passed/failed/skipped counts, duration, failed tests with full suite > test paths and error messages, and retried tests. A green 350-test run is now ~5 lines.
• Actionable error diagnostics (#10) — known infrastructure failures print cause + remediation instead of raw stack traces: dev server unreachable, TWD sidebar not mounting, Puppeteer protocolTimeout, and Chrome launch failures. The CLI can no longer exit 1 silently.

Breaking (output only)

The old Tests: N passed… line and per-test tree are gone — update any log-parsers. Exit codes are unchanged (0 pass / 1 failure), so scripts and CI gates keep working.

Dependencies

• Removed twd-js from dependencies (no longer used at runtime — the tested app provides it)

CI

• GitHub prereleases now publish to npm under the beta dist-tag

Full Changelog: v1.2.0...v1.3.0

v1.3.0-beta.1 — AI-friendly output (beta)

Beta pre-release of the AI-friendly output rework (PR #10) for testing in real projects before merge.

What's new

• Relay-style --- Run complete --- summary block, always the last output: passed/failed/skipped counts, duration, failed tests with suite > test paths and error messages, retried tests
• Actionable diagnostics for known infrastructure failures (dev server down, TWD sidebar missing, protocolTimeout, Chrome launch failure) instead of raw stack traces
• Removed noise: config JSON dump, per-test tree, status chatter — green runs are ~5 lines
• The CLI can no longer exit 1 silently
• twd-js removed from dependencies (no longer used)

Breaking (output only)

The old Tests: N passed… line and per-test tree are gone — update any log-parsers. Exit codes are unchanged (0 pass / 1 failure).

Install

npm install -D twd-cli@beta

🤖 Generated with Claude Code

v1.2.0

What's Changed

Features

• Add `--test` filter flag for targeted test runs (#9) — run a specific test (or subset) instead of the full suite.

Fixes

• Regenerate lockfile on Linux to sync `@emnapi` to 1.10.0, and add `npm run lock:linux` for future dependency updates.

Dependencies

• Update `twd-js` to 1.8.1
• Update `puppeteer` to 25.1.0
• Update `vitest` to 4.1.8

Full Changelog: v1.1.15...v1.2.0

v1.1.15

• chore: update twd-js to 1.8.1
• chore: update puppeteer to 25.1.0
• chore: update vitest to 4.1.8
• fix: regenerate lockfile on linux to sync @emnapi to 1.10.0 (adds npm run lock:linux for future dependency updates)

v1.1.14

• feat: configurable protocolTimeout to avoid 180s suite aborts (#8)

The whole test suite runs inside a single page.evaluate, so Puppeteer's implicit 180s protocolTimeout could abort long-but-passing suites on slow CI with no per-test output. Adds a configurable protocolTimeout (default 300000 / 5 min, 0 = no timeout) and a clearer error hint when a run aborts on timeout.

v1.1.13

• feat: add Tests: summary line, Failed tests block, and MOCK prefix on contract lines (#7)

v1.1.12

• chore: update twd-js to 1.8.0
• chore: update puppeteer to 24.43.0
• chore: update openapi-mock-validator to 0.3.0

v1.1.11

• chore: update dependencies

v1.1.10

• fix: forward mock Content-Type to contract validator (avoids false-positive MISSING_SCHEMA for binary mocks) (#5)