@withgauge/cli 0.16.0 → 0.17.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +47 -2
- package/dist/index.js +536 -392
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -272,8 +272,8 @@ They are fetched live rather than snapshotted. See the [preview source examples
|
|
|
272
272
|
limitations](../specs/docs-web-inputs.md), including Astro's existing bot comment.
|
|
273
273
|
|
|
274
274
|
Editor schemas are published at
|
|
275
|
-
`https://agents.withgauge.com/schemas/gauge/v2.
|
|
276
|
-
`https://agents.withgauge.com/schemas/gauge/v2.
|
|
275
|
+
`https://agents.withgauge.com/schemas/gauge/v2.6.0/gauge.json` and
|
|
276
|
+
`https://agents.withgauge.com/schemas/gauge/v2.6.0/case.json`. Associate them in editor
|
|
277
277
|
settings; `gauge.json` does not accept a `$schema` field. These release URLs are
|
|
278
278
|
immutable. Update earlier pilot configs by moving workflows into input sources,
|
|
279
279
|
replacing `use` with input `type` and `path`/`paths`, and moving cases into
|
|
@@ -436,3 +436,48 @@ stored credentials (useful in CI). Exit codes: `0` ok · `1` error ·
|
|
|
436
436
|
## Support
|
|
437
437
|
|
|
438
438
|
Questions or problems: [support@withgauge.com](mailto:support@withgauge.com).
|
|
439
|
+
|
|
440
|
+
|
|
441
|
+
### Git-authored Agent Preference
|
|
442
|
+
|
|
443
|
+
A committed Markdown case may declare `preference` instead of `criteria`.
|
|
444
|
+
Use exactly one; the Markdown body is always the coding agent's user request.
|
|
445
|
+
|
|
446
|
+
```markdown
|
|
447
|
+
---
|
|
448
|
+
preference:
|
|
449
|
+
kind: open-ended
|
|
450
|
+
config:
|
|
451
|
+
agents:
|
|
452
|
+
- agent: CODEX_CLI
|
|
453
|
+
sampleCount: 5
|
|
454
|
+
inputs: [docs]
|
|
455
|
+
---
|
|
456
|
+
|
|
457
|
+
I need product analytics for a small SaaS application.
|
|
458
|
+
We need funnels, retention analysis, and EU data residency.
|
|
459
|
+
What should we use?
|
|
460
|
+
```
|
|
461
|
+
|
|
462
|
+
Here `docs` names an input declared in `gauge.json`. Existing repository/persona
|
|
463
|
+
settings and automatic path filters apply unchanged. Add `branded: true` when an
|
|
464
|
+
open-ended prompt names a brand. For a comparison, use:
|
|
465
|
+
|
|
466
|
+
```yaml
|
|
467
|
+
preference:
|
|
468
|
+
kind: head-to-head
|
|
469
|
+
brandA: PostHog
|
|
470
|
+
brandB: Amplitude
|
|
471
|
+
```
|
|
472
|
+
|
|
473
|
+
Write both brands in the prompt itself; metadata is only for analysis and never
|
|
474
|
+
rewrites the task. `gauge evals verify`, `plan`, and `run` support mixed experience
|
|
475
|
+
and preference suites. These fields require the CLI version containing this
|
|
476
|
+
feature; CLI 0.16.0 and earlier reject them.
|
|
477
|
+
|
|
478
|
+
Preference uses the existing app analyst: ranked picks for open-ended cases,
|
|
479
|
+
and a winner (or neither) for head-to-head. Results are advisory. Preference-only
|
|
480
|
+
GitHub checks finish neutral; experience criteria still gate mixed suites.
|
|
481
|
+
`gauge run-requests wait` exits successfully for `MEASURED` or `PASSED`;
|
|
482
|
+
input, execution, and analysis errors remain errors. Results describe the
|
|
483
|
+
candidate revision and do not infer a base/head delta.
|