> ## Documentation Index > Fetch the complete documentation index at: https://docs.cekura.ai/llms.txt > Use this file to discover all available pages before exploring further. # Get Combined Results Report > Aggregate two or more test results into a single report with success rate, metric summary, critical categories, latency and failure reasons ## OpenAPI ````yaml get /test_framework/v2/results/reports/ openapi: 3.1.0 info: title: Cekura API version: v1 description: >- Complete API documentation for the Cekura platform. This API provides endpoints for testing, observing, and evaluating AI voice agents — including managing agents, running evaluators, defining metrics, and analyzing call quality. servers: - url: https://api.cekura.ai security: [] paths: /test_framework/v2/results/reports/: get: tags: - Results summary: Get a combined report across results description: >- Aggregate two or more test results into a single report: overall success rate, per-metric summary, run counts, critical categories, per-tag success rates, response-latency percentiles, deduplicated failure reasons and combined cost/duration figures. Find the result IDs with `results_list` first. Passing a single ID is allowed but `results_retrieve` is the better tool for one result. Leave `ql` at its default `{-runs}`. Selecting runs embeds every run of every result in the response; fetch the runs you actually need with `runs_bulk_retrieve` instead. operationId: results-reports-retrieve_3 parameters: - in: query name: result_ids schema: type: string description: >- Results to combine — comma-separated string (e.g. '5591,5592'), not a JSON array. required: true - in: query name: run_filters schema: type: string description: > Optional JSON-encoded filter object restricting which runs the report counts. Every aggregate in the response is computed over the matching runs only. Invalid JSON returns 400. `join` is `and` (every filter must match) or `or` (any filter may match). Each item in `filters` has an `id` (the run field), an `operator` and a `value`. Supported filters: - `success` — `equals` with `true` or `false` - `status` — `equals` with one run status, e.g. `"completed"` - `id` — `equals` or `in` with run IDs - `scenarios` — `in` with scenario IDs - `tags` — `in` with scenario tags; a run matches if its scenario carries any of them - `scenario_name`, `call_ended_reason` — `in` with a list of values - `metrics` — no top-level `operator`; `value` is a list of metric conditions, each with the metric `id` and an `operator`: `equals` (with `value`), `greaterThanOrEqual` / `lessThanOrEqual` (with `value`), `range` (with `min` and `max`), or `exists` / `notExists` Example — failed runs from two scenarios whose metric 101 scored at least 3: ```json { "join": "and", "filters": [ {"id": "success", "operator": "equals", "value": false}, {"id": "scenarios", "operator": "in", "value": [3120, 3121]}, {"id": "metrics", "value": [{"id": 101, "operator": "greaterThanOrEqual", "value": 3}]} ] } ``` - in: query name: ql schema: type: string default: '{-runs}' description: >- Field selection. `{-runs}` (the default) returns the aggregates only; `{*}` additionally embeds every run and can be very large. - name: agent_id in: query required: false schema: type: integer description: >- Tenant scope. Supply one of `agent_id`, `assistant_id`, `project_id` when authenticating with a user session or OAuth bearer token; omit when using an API key already scoped to the tenant. - name: assistant_id in: query required: false schema: type: string description: >- Tenant scope. Supply one of `agent_id`, `assistant_id`, `project_id` when authenticating with a user session or OAuth bearer token; omit when using an API key already scoped to the tenant. - name: project_id in: query required: false schema: type: integer description: >- Tenant scope. Supply one of `agent_id`, `assistant_id`, `project_id` when authenticating with a user session or OAuth bearer token; omit when using an API key already scoped to the tenant. responses: '200': content: application/json: schema: $ref: '#/components/schemas/SchemaCombinedResultsReport' description: '' '400': content: application/json: schema: type: object properties: field_name: type: array items: type: string description: '' security: - api_key: [] - oauth2: [] - embedded_session: [] - shareable_link: [] - supabase_session: [] components: schemas: SchemaCombinedResultsReport: type: object properties: status: type: string description: Always 'completed' — the report is computed on read, not queued. success_rate: type: number format: double description: Percentage of runs that passed, across every result in the report total_runs_count: type: integer description: Runs in the report, including those still executing completed_runs_count: type: integer description: >- Runs that reached a terminal state (completed, failed, timeout or cancelled) success_runs_count: type: integer description: Completed runs that passed failed_runs_count: type: integer description: Completed runs that failed agent: type: integer description: ID of the agent the results belong to agent_version: oneOf: - {} - type: 'null' description: > Agent version every result ran against, or `null` when the results did not all run against the same version. Example: ```json { "history_id": 812, "version_id": "v3", "version_name": "Pricing prompt fix", "is_current": true } ``` overall_evaluation: description: > Aggregate evaluation across all runs in the report. - `success_rate` — percentage of runs that passed - `metric_summary` — keyed by metric ID. Every entry has `id`, `name`, `type`, `score`, `errors` (the scenarios whose runs failed the metric, with the failing run IDs) and `overall_explanation`. Enum metrics add `enums` (per enum value: `score` as a percentage, `count` and `runs`), numeric metrics add percentiles such as `p50`, and critical metrics add `categories` keyed by category ID - `worst_performing_metrics` — `binary_adherence` lists the IDs of binary metrics that did not pass on every run, lowest pass rate first - `numeric_metrics` — numeric metric averages with their percentiles, plus average ringing duration when available - `enum_metrics` — the metric ID of every enum evaluation, one entry per evaluated run - `extra_metrics` — derived figures such as average ringing duration and expected-outcome score Example: ```json { "success_rate": 66.67, "metric_summary": { "101": { "id": 101, "name": "Politeness", "type": "binary", "vocera_defined_metric_code": "", "score": 0.67, "overall_explanation": [], "errors": [ { "id": 3120, "name": "Asks for pricing", "agent": 42, "personality": 7, "personality_name": "Busy professional", "retell_agent_id": "", "instructions": "", "is_simulating": false, "runs": [90012] } ] }, "102": { "id": 102, "name": "Sentiment", "type": "enum", "vocera_defined_metric_code": "", "score": null, "overall_explanation": [], "errors": [], "enums": { "positive": {"score": 100.0, "count": 3, "runs": [{"id": 90011}, {"id": 90012}, {"id": 90013}]} } }, "103": { "id": 103, "name": "Latency", "type": "numeric", "vocera_defined_metric_code": "", "score": 850.0, "p50": 800.0, "overall_explanation": [], "errors": [] }, "104": { "id": 104, "name": "Critical Deviations", "type": "binary", "vocera_defined_metric_code": "", "score": 0.67, "overall_explanation": [], "errors": [], "categories": { "55": { "id": 55, "category": "Wrong price quoted", "runs": [{"id": 90012}], "runs_count": 1, "priority_counts": {"critical": 1, "non-critical": 0, "not_a_bug": 0}, "scenarios": {} } } } }, "worst_performing_metrics": {"binary_adherence": [101, 104]}, "numeric_metrics": [ {"name": "Latency", "type": "numeric", "value": 850.0, "percentiles": {"p50": 800.0}}, {"name": "Average Ringing Duration", "type": "numeric", "value": 2.5, "percentiles": {"p50": 2.5}} ], "enum_metrics": [102, 102, 102], "extra_metrics": [ {"name": "Average Ringing Duration", "type": "numeric", "value": 2.5, "percentiles": {"p50": 2.5}}, {"name": "Expected Outcome", "type": "numeric", "value": 3.33} ] } ``` metrics: description: | Metrics configured on the agent. Example: ```json [ { "id": 101, "name": "Politeness", "eval_type": "binary", "simulation_enabled": true, "observability_enabled": true, "function_name": null, "vocera_defined_metric_code": "" } ] ``` critical_categories: description: > Critical-issue categories found across the runs, most critical first. Categories marked as not a bug and evaluations marked as wrong are excluded. Example: ```json [ { "id": 55, "category": "Wrong price quoted", "scenarios": {}, "runs_count": 1, "runs": [{"id": 90012}], "priority_counts": {"critical": 1, "non-critical": 0, "not_a_bug": 0} } ] ``` runs_by_tags: description: > Keyed by scenario tag: the IDs of the runs whose scenario carries the tag and the success rate (percentage) of those runs. Example: ```json { "billing": {"runs": [90013, 90012, 90011], "success_rate": 66.67} } ``` latency_data: description: > Response latency in milliseconds across all measured turns of all runs. Example: ```json {"mean": 800.0, "median": 800.0, "p50": 800.0, "p95": 890.0, "p99": 898.0} ``` failed_reasons: description: > Failure issues merged and deduplicated across the results, ranked by how many runs they affect. Example: ```json { "issues": [ { "rank": 1, "title": "Missed pricing", "description": "The agent never shared pricing when asked.", "run_ids": [90012], "affected_count": 1 } ], "total_failed_runs": 1 } ``` performance_metrics: description: > Per-metric performance across the runs, rubric metrics first. `aggregate_unit` is `percent`, `value`, `score` or an empty string. `passed` and `threshold_text` come from the project rubric; `passed` is `null` and `in_rubric` is `false` for metrics without a rubric rule. Enum metrics add `enum_run_ids` (run IDs per enum value), rating metrics add `score_range_run_ids` and expected-outcome metrics add `review_required_run_ids`. Example: ```json [ { "metric_id": 101, "metric_name": "Politeness", "eval_type": "binary", "aggregate_value": 66.7, "aggregate_unit": "percent", "passed": false, "threshold_text": ">= 80%", "evaluated_runs_count": 3, "evaluated_run_ids": [90013, 90012, 90011], "failing_run_ids": [90012], "passing_run_ids": [90013, 90011], "in_rubric": true }, { "metric_id": 105, "metric_name": "Clarity", "eval_type": "continuous_qualitative", "aggregate_value": 4.0, "aggregate_unit": "score", "passed": null, "threshold_text": "", "evaluated_runs_count": 3, "evaluated_run_ids": [90013, 90012, 90011], "failing_run_ids": [], "passing_run_ids": [], "in_rubric": false, "score_range_run_ids": {"0-50": [], "50-75": [], ">75": [90013, 90012, 90011]} } ] ``` runs: description: > Every run in the report, keyed by run ID. Present only when `ql` selects it (e.g. `{*}`); the default `{-runs}` omits it. Example: ```json { "90012": { "id": 90012, "result_id": 5591, "result_name": "Nightly regression", "scenario": 3120, "scenario_name": "Asks for pricing", "personality_name": "Busy professional", "test_profile_name": null, "expected_outcome": {"score": 0}, "success": false, "status": "completed", "evaluation_status": "failure", "evaluation": { "metrics": [ { "id": 101, "name": "Politeness", "type": "binary", "score": false, "score_normalized": 0, "explanation": ["The agent interrupted the caller twice."], "extra": {}, "vocera_defined_metric_code": "", "function_name": null, "thumbs_up": null }, { "id": 102, "name": "Sentiment", "type": "enum", "enum": "positive", "explanation": [], "extra": {}, "vocera_defined_metric_code": "", "function_name": null, "thumbs_up": null } ] }, "timestamp": "2026-09-24T08:43:10.473761Z", "executed_at": "2026-09-24T08:43:12.000000Z", "duration": "01:00", "error_message": "", "metadata": {"ringing_duration": 2.5}, "inbound_number": "", "outbound_dial_window_opens_at": null, "outbound_dial_window_seconds": null, "outbound_dial_window_closes_at": null } } ``` required: - agent - agent_version - completed_runs_count - critical_categories - failed_reasons - failed_runs_count - latency_data - metrics - overall_evaluation - performance_metrics - runs_by_tags - status - success_rate - success_runs_count - total_runs_count securitySchemes: api_key: type: apiKey in: header name: X-CEKURA-API-KEY description: >- API Key Authentication. It should be included in the header of each request. oauth2: type: http scheme: bearer bearerFormat: JWT description: OAuth access token issued by Cekura for connected apps. embedded_session: type: http scheme: bearer bearerFormat: JWT description: Embedded-dashboard session token. Internal. shareable_link: type: apiKey in: header name: X-VOCERA-SHAREABLE-LINK-TOKEN description: Share-link token scoped to a single shared resource. supabase_session: type: http scheme: bearer bearerFormat: JWT description: Cekura dashboard session token. Not a customer API credential. ````