woobe-cli 0.24.0 → 0.25.0
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/USAGE.md
CHANGED
|
@@ -770,3 +770,47 @@ not uploaded. Acceptance is asynchronous preparation, without publication or
|
|
|
770
770
|
environment selection. Lost responses use `draft reconcile`, preserving the
|
|
771
771
|
original operation. See [ASaC](ASAC.md#prepare-an-exact-immutable-candidate) for
|
|
772
772
|
checkpoint prerequisites, status, bindings and recovery.
|
|
773
|
+
|
|
774
|
+
## Exact Candidate evaluation
|
|
775
|
+
|
|
776
|
+
After an isolated `stage --revision`, inspect the Candidate until its state is
|
|
777
|
+
`ready`. Evaluate its UUID with a versioned YAML suite; current local YAML and
|
|
778
|
+
native environment selections cannot replace this frozen snapshot.
|
|
779
|
+
|
|
780
|
+
```yaml
|
|
781
|
+
schema_version: '1.0'
|
|
782
|
+
suite_id: support-smoke
|
|
783
|
+
suite_version: '1'
|
|
784
|
+
dataset_version: '1'
|
|
785
|
+
policy_version: all-cases-pass@1
|
|
786
|
+
cases:
|
|
787
|
+
- id: greeting
|
|
788
|
+
message: Hello
|
|
789
|
+
# Optional: expected_output is compared to actual structured output,
|
|
790
|
+
# or {answer: ...} for unstructured text. Omit it for execution/schema checks.
|
|
791
|
+
```
|
|
792
|
+
|
|
793
|
+
```sh
|
|
794
|
+
woobe agent '@support' test --candidate CANDIDATE_UUID --file suite.yaml --yes
|
|
795
|
+
woobe network '@helpdesk' test --candidate CANDIDATE_UUID --file suite.yaml --yes
|
|
796
|
+
woobe agent '@support' evaluation EVALUATION_UUID
|
|
797
|
+
woobe network '@helpdesk' evaluation EVALUATION_UUID reconcile --yes
|
|
798
|
+
```
|
|
799
|
+
|
|
800
|
+
The suite accepts 1–16 uniquely named cases. Suite, dataset and policy versions
|
|
801
|
+
are declared identities; the backend separately verifies content digests. Every
|
|
802
|
+
case gets its original native Run identity before execution. Qualification uses
|
|
803
|
+
actual terminal Run evidence, exact snapshot identity, output validation and
|
|
804
|
+
current binding fingerprints. Missing evidence stays incomplete. Failed cases
|
|
805
|
+
exit with code 6; `accepted` and `running` do not mean passed.
|
|
806
|
+
|
|
807
|
+
The immutable acceptance receipt identifies the evaluation. Inspect its ledger
|
|
808
|
+
for the current result. If acceptance is uncertain, use `draft reconcile` before
|
|
809
|
+
another write: it recovers the original receipt and never executes a case again.
|
|
810
|
+
The evaluation's own `reconcile` reads original Run evidence without provider
|
|
811
|
+
calls. Neither command publishes nor changes Production. Do not combine
|
|
812
|
+
`test --candidate` with `--env` or `--version`. Native Agent `test --env` remains
|
|
813
|
+
available; Network testing requires an exact Candidate. Inspection by native UUID
|
|
814
|
+
works without `.woobe-config`; accepting an evaluation uses private project state
|
|
815
|
+
to preserve its write identity. Exact Candidate evaluation requires CLI 0.25+
|
|
816
|
+
and a backend advertising `asac.candidate_evaluation`.
|
package/package.json
CHANGED
|
Binary file
|
|
Binary file
|
package/vendor/linux-amd64/woobe
CHANGED
|
Binary file
|
package/vendor/linux-arm64/woobe
CHANGED
|
Binary file
|
|
Binary file
|
|
Binary file
|