Skip to content

v1.2.1

Choose a tag to compare

View all tags
@kevinccbsg kevinccbsg released this 25 May 14:55
v1.2.1
5919618

Patch release tightening twd-relay run output for AI-agent consumers.

Improvement

twd-relay run is now silent during the run — summary only (#10)

The CLI no longer streams per-test RUN: / PASS: / SKIP: lines or inline FAIL: lines, and the connect chatter is trimmed. A green run now looks like:

Connected to ws://localhost:5173/__twd/ws
Running 16 test(s)...

--- Run complete ---
  Passed: 16 | Failed: 0 | Skipped: 0
  Duration: 5.3s

A run with failures looks like:

Connected to ws://localhost:5173/__twd/ws
Running 16 test(s)...

--- Run complete ---
  Passed: 15 | Failed: 1 | Skipped: 0
  Duration: 5.3s

  Failed tests (1):
    × Cars Delete > should open the confirm dialog, delete the car, and remove the row from the list
      Assertion failed: Expected text to contain "with i1", but got "ConfirmAre you sure you want to delete this item? with id 1? ..."

Why: The primary consumer of twd-relay run output is now the twd Claude Code skill, which feeds stdout straight into the model's context. The old streaming format printed ~32 per-test lines per run that the AI then had to grep through for failures. The new format is a single end-of-run block — same failure information, far fewer tokens. Long float durations like 625.7999999523163ms are gone with the per-test lines.

No API changes. Wire protocol, exit codes, and CLI flags (--test, --port, --path, --host, --timeout, --max-test-duration) are all unchanged. Humans watching a slow run who want live progress can use the in-browser TWD sidebar.

Full Changelog: v1.2.0...v1.2.1

Assets 2
Loading