[COVAL-7144] Reconcile the CLI with the Oct 5 API catalog and release 0.9.0

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "coval"
-version = "0.8.3"
+version = "0.9.0"
 edition = "2021"
 description = "CLI for Coval AI agent evaluation platform"
 license = "MIT"

api-coverage-report.md

@@ -13,13 +13,13 @@ only when coverage actually changes.
 | Metric | Value |
 | --- | ---: |
 | Reconciliation status | PASS |
-| Published operations | 181 |
+| Published operations | 183 |
 | First-class CLI operations | 119 |
-| Reviewed gaps | 62 |
+| Reviewed gaps | 64 |
 | Client operations | 143 |
-| Published request fields on covered operations | 365 |
-| Request fields modeled by the CLI | 359 |
-| Reviewed request-field gaps | 6 |
+| Published request fields on covered operations | 437 |
+| Request fields modeled by the CLI | 437 |
+| Reviewed request-field gaps | 0 |
 
 Catalog: https://api.coval.dev/v1/openapi
 
@@ -112,6 +112,8 @@ Catalog: https://api.coval.dev/v1/openapi
 - `POST /personas/{persona_id}/versions/{version_id}/revert`
 - `POST /review-annotations:withMetricOutputs`
 - `POST /review-projects/disagreement-state`
+- `POST /review-projects/{project_id}/complete-conversation`
+- `POST /sofia/delegation-token`
 - `POST /test-sets/{test_set_id}/agents:add`
 - `POST /test-sets/{test_set_id}/duplicate`
 - `POST /test-sets/{test_set_id}/versions/{version_id}/revert`
@@ -138,9 +140,4 @@ Catalog: https://api.coval.dev/v1/openapi
 
 ## All current request-field gaps
 
-- `PATCH /run-templates/{id} agent_id`
-- `PATCH /run-templates/{id} persona_id`
-- `PATCH /run-templates/{id} test_set_id`
-- `POST /run-templates agent_id`
-- `POST /run-templates persona_id`
-- `POST /run-templates test_set_id`
+- None.

api-coverage.toml

@@ -5,11 +5,11 @@
 
 [snapshot]
 catalog_url = "https://api.coval.dev/v1/openapi"
-reviewed_at = "2026-09-14"
-published_operations = 181
+reviewed_at = "2026-10-05"
+published_operations = 183
 cli_supported_operations = 119
-published_request_fields = 365
-cli_modeled_request_fields = 359
+published_request_fields = 437
+cli_modeled_request_fields = 437
 
 [[allowed_extra]]
 operation = "POST /test-cases/{test_case_id}/media:upload-url"
@@ -376,6 +376,14 @@ reason = "Workspace management remains to be modeled under COVAL-2079."
 operation = "POST /workspaces/{workspace_id}/archive"
 reason = "Workspace management remains to be modeled under COVAL-2079."
 
+[[known_gap]]
+operation = "POST /review-projects/{project_id}/complete-conversation"
+reason = "Project-scoped completion remains to be modeled under COVAL-2079; this explicit write needs a deliberate command workflow rather than being exposed by the parity refresh."
+
+[[known_gap]]
+operation = "POST /sofia/delegation-token"
+reason = "Customer API-key delegation token minting remains to be modeled under COVAL-2079 with an explicit security-sensitive CLI workflow."
+
 # =============================================================================
 # REQUEST FIELDS
 # =============================================================================
@@ -391,36 +399,6 @@ reason = "Workspace management remains to be modeled under COVAL-2079."
 # Pydantic model; coval-ai/backend records the known divergence in
 # src/services/api/tests/v1/openapi_parity_baseline.txt.
 
-[[known_field_gap]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "agent_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `agent_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
-[[known_field_gap]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "persona_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `persona_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
-[[known_field_gap]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "test_set_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `test_set_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
-[[known_field_gap]]
-operation = "POST /run-templates"
-field = "agent_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `agent_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
-[[known_field_gap]]
-operation = "POST /run-templates"
-field = "persona_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `persona_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
-[[known_field_gap]]
-operation = "POST /run-templates"
-field = "test_set_id"
-reason = "Documented but not served: the transport forbids extra fields and takes the plural `test_set_ids` array the CLI already sends, so modeling this name would break the call. Spec drift tracked under COVAL-5825."
-
 # Fields the CLI sends that the published schema does not declare. Most are served
 # but undocumented, so removing them from the CLI would lose working behavior.
 
@@ -429,21 +407,6 @@ operation = "PATCH /conversations/uploaded/{conversation_id}"
 field = "audio_reference"
 reason = "Served but deliberately hidden from the published schema as an internal field; the API still accepts it."
 
-[[allowed_extra_field]]
-operation = "PATCH /metrics/{metric_id}"
-field = "case_insensitive"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
-[[allowed_extra_field]]
-operation = "PATCH /metrics/{metric_id}"
-field = "match_mode"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
-[[allowed_extra_field]]
-operation = "PATCH /metrics/{metric_id}"
-field = "position"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
 [[allowed_extra_field]]
 operation = "PATCH /review-annotations/{annotation_id}"
 field = "completion_status"
@@ -454,26 +417,6 @@ operation = "PATCH /review-annotations/{annotation_id}"
 field = "status"
 reason = "Advertised by the CLI but absent from the served contract, which ignores unknown fields, so the flag silently does nothing. Tracked under COVAL-5824."
 
-[[allowed_extra_field]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "agent_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."
-
-[[allowed_extra_field]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "persona_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."
-
-[[allowed_extra_field]]
-operation = "PATCH /run-templates/{run_template_id}"
-field = "test_set_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."
-
-[[allowed_extra_field]]
-operation = "POST /api-keys"
-field = "environment"
-reason = "Served but undocumented; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
 [[allowed_extra_field]]
 operation = "POST /conversations/simulated/{simulation_id}/resimulate"
 field = "dev_id"
@@ -483,33 +426,3 @@ reason = "Served but deliberately hidden from the published schema as an interna
 operation = "POST /conversations/uploaded:submit"
 field = "audio"
 reason = "Served but deliberately hidden from the published schema as an internal field; the API still accepts it."
-
-[[allowed_extra_field]]
-operation = "POST /metrics"
-field = "case_insensitive"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
-[[allowed_extra_field]]
-operation = "POST /metrics"
-field = "match_mode"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
-[[allowed_extra_field]]
-operation = "POST /metrics"
-field = "position"
-reason = "Served but undocumented regex-matching field; recorded as spec drift in the backend's own openapi_parity_baseline.txt."
-
-[[allowed_extra_field]]
-operation = "POST /run-templates"
-field = "agent_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."
-
-[[allowed_extra_field]]
-operation = "POST /run-templates"
-field = "persona_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."
-
-[[allowed_extra_field]]
-operation = "POST /run-templates"
-field = "test_set_ids"
-reason = "Served but undocumented: the run-template models take these plural arrays. The published singular form is stale spec, tracked under COVAL-5825."

src/client/models/metric.rs

@@ -13,6 +13,9 @@ pub struct Metric {
     pub metric_type: MetricType,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub prompt: Option<String>,
+    /// Agent Judge evidence tools. An empty list disables all tools.
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub enabled_tools: Option<Vec<String>>,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub categories: Option<Vec<String>>,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -116,6 +119,47 @@ pub struct CreateMetricRequest {
     pub metric_type: MetricType,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub prompt: Option<String>,
+    /// Agent Judge evidence tools. An empty list disables all tools.
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub enabled_tools: Option<Vec<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub judge_mode: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub aggregation_method: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub unit: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub detection_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub harmonics_to_noise_ratio_threshold_offset_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub jitter_threshold_multiplier: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub loud_threshold_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub low_pitch_threshold_multiplier: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub mad_z_score_threshold: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub metric_attribute: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub metric_metadata: Option<serde_json::Value>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub min_fry_segment_seconds: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub pause_detection_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub pitch_change_threshold_hz: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub significant_changes_threshold_hz: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub soft_threshold_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub span_name: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub threshold_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub value_source: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub categories: Option<Vec<String>>,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -202,6 +246,57 @@ pub struct UpdateMetricRequest {
     pub metric_type: Option<MetricType>,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub prompt: Option<String>,
+    /// Agent Judge evidence tools. An empty list disables all tools.
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub enabled_tools: Option<Vec<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub judge_mode: Option<String>,
+    /// SQL and custom-trace aggregation; an explicit null resets it to the default.
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub aggregation_method: Option<Option<String>>,
+    /// Display unit; an explicit null clears it.
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub unit: Option<Option<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub detection_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub harmonics_to_noise_ratio_threshold_offset_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub jitter_threshold_multiplier: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub loud_threshold_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub low_pitch_threshold_multiplier: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub mad_z_score_threshold: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub metric_attribute: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub metric_metadata: Option<serde_json::Value>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub min_fry_segment_seconds: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub pause_detection_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub pitch_change_threshold_hz: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub significant_changes_threshold_hz: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub soft_threshold_db: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub span_name: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub threshold_preset: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub value_source: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub categories: Option<Vec<String>>,
     #[serde(skip_serializing_if = "Option::is_none")]

src/client/models/persona.rs

@@ -97,6 +97,18 @@ pub struct CreatePersonaRequest {
     /// `situate_speaker`.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub audio_degradation: Option<serde_json::Value>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub silent_mode: Option<bool>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub multi_phone_config: Option<serde_json::Value>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub initialization_parameters: Option<serde_json::Value>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub custom_persona_data: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub voice: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub custom_voice_id: Option<String>,
     /// Tag names. Omitted leaves tags unchanged; an empty list clears them.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub tags: Option<Vec<String>>,
@@ -158,6 +170,44 @@ pub struct UpdatePersonaRequest {
         skip_serializing_if = "Option::is_none"
     )]
     pub audio_degradation: Option<Option<serde_json::Value>>,
+    // Advanced persona values are stored in metadata. The API uses field
+    // presence to distinguish an omitted update from an explicit null clear.
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub silent_mode: Option<Option<bool>>,
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub multi_phone_config: Option<Option<serde_json::Value>>,
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub initialization_parameters: Option<Option<serde_json::Value>>,
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub custom_persona_data: Option<Option<String>>,
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub voice: Option<Option<String>>,
+    #[serde(
+        default,
+        deserialize_with = "super::explicit_option",
+        skip_serializing_if = "Option::is_none"
+    )]
+    pub custom_voice_id: Option<Option<String>>,
     /// Tag names. Omitted leaves tags unchanged; an empty list clears them.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub tags: Option<Vec<String>>,