Test outputs (results, recommendations, report)

After a Redraven test is evaluated, you can read metrics, policy recommendations, and the PDF report over HTTP. These endpoints do not return per-case prompts or full LLM transcripts.

See Authentication for headers and base URL. Replace {test_id} with your test UUID.

Pipeline vs metrics API

The Python SDK eval flow uses GET /tests/{id}/results/summary?kind=eval and GET /tests/{id}/results/cases for pipeline materialization. The endpoints on this page are the stable read APIs for dashboard metrics, recommendation export, and report download.

Prerequisites

Output	Requirement
Results overview	Test exists in your org; evaluation finished (status is completed or evaluation_failed)
Recommendations	Rows stored in policy_recommendations (created after evaluation, or via POST /tests/{id}/recommendations/generate)
Report PDF	Report generated (POST /tests/{id}/report/generate-reportlab or Generate Report in the app), then GET /tests/{id}/report/status shows ready: true

---

Test results overview

GET {base}/api/v1/tests/{test_id}/results

Returns pass/fail totals and pass rates grouped by certification and policy. No request/response text for individual test cases.

When to use

• Export or integrate the same metrics shown on the test results page in the Redraven app
• Automate compliance dashboards without pulling case-level artifacts

Request

GET test results overview

curl -sS \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/results"

Response

Field	Type	Description
test_id	string (UUID)	Test identifier
created_at	string (ISO 8601)	Test creation time
status	string or null	Test status (e.g. completed)
summary	object	Overall total, passed, failed, pass_rate
certifications	array	Groups per certification, each with summary and policies[]

pass_rate on this endpoint

pass_rate values are fractions between 0.0 and 1.0 (for example 0.85 means 85%). This differs from raw aggregated_policies storage (integer 0–100) and from recommendation pass_rate (see below).

Example response

Results overview (trimmed)

{
  "test_id": "11111111-1111-1111-1111-111111111111",
  "created_at": "2025-06-01T12:00:00Z",
  "status": "completed",
  "summary": {
    "total": 100,
    "passed": 85,
    "failed": 15,
    "pass_rate": 0.85
  },
  "certifications": [
    {
      "certification": "EU AI Act",
      "summary": {
        "total": 60,
        "passed": 50,
        "failed": 10,
        "pass_rate": 0.8333333333333334
      },
      "policies": [
        {
          "name": "Policy A",
          "total": 30,
          "passed": 28,
          "failed": 2,
          "pass_rate": 0.9333333333333333
        }
      ]
    }
  ]
}

Errors

• 404 — Test not found, or evaluation not finished / no aggregate data yet (detail explains wait for evaluation)

---

Recommendations

GET {base}/api/v1/tests/{test_id}/recommendations

Returns stored FireGuard-style policy recommendations generated from failing policies on the test.

Query parameters

Parameter	Default	Description
limit	10	Maximum number of rows returned (UI-style list)
limit	0	Return all stored recommendations (export)
sort	pass_rate_asc	pass_rate_asc or pass_rate_desc

When limit is greater than 0, it is capped at 500.

Request (UI list)

GET top recommendations

curl -sS \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/recommendations?limit=10&sort=pass_rate_asc"

Request (export all)

GET all recommendations

curl -sS \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/recommendations?limit=0&sort=pass_rate_asc"

Response

Field	Type	Description
recommendations	array	Recommendation objects
count	integer	Length of recommendations

Each recommendation includes id, redraven_policy_id, redraven_policy_text, certification, pass_rate, test_cases_total, test_cases_failed, fireguard_policy, created_at.

pass_rate on recommendations

pass_rate here is stored in database format (typically 0–100 as a percentage, e.g. 42 for 42%). It is not the 0.0–1.0 fraction used on the results overview endpoint.

Example response

Recommendations (trimmed)

{
  "recommendations": [
    {
      "id": "22222222-2222-2222-2222-222222222222",
      "redraven_policy_id": "policy-hash-id",
      "redraven_policy_text": "Do not disclose internal credentials.",
      "certification": "EU AI Act",
      "pass_rate": 42,
      "test_cases_total": 12,
      "test_cases_failed": 7,
      "fireguard_policy": {
        "name": "Block credential disclosure",
        "description": "Reject requests that ask for secrets.",
        "criticality": "HIGH",
        "violation": true,
        "metadata": {}
      },
      "created_at": "2025-06-01T14:00:00Z"
    }
  ],
  "count": 1
}

---

Report download

The report is a PDF file, not JSON. Generate it first, poll status, then download bytes.

Step 1 — Generate report

POST generate report (ReportLab)

curl -sS -X POST \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/report/generate-reportlab"

You can also use Generate Report on the test results page in the Redraven app.

Step 2 — Poll until ready

GET report status

curl -sS \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/report/status"

Proceed when ready is true (and optionally file_id is set).

Step 3 — Download PDF

GET report download

curl -sS -o report.pdf \
  -H "X-API-Key: ${REDRAVEN_API_KEY}" \
  -H "X-Organization-Id: ${REDRAVEN_ORGANIZATION_ID}" \
  "${REDRAVEN_BASE_URL}/api/v1/tests/{test_id}/report/download"

The response body is binary PDF data with Content-Disposition: attachment. Do not parse it as JSON.

Errors

• 404 — Report not generated or file missing in storage

---

End-to-end flow

1. Run a test to completion (SDK generate_and_run_test, or the app UI).
2. GET /tests/{test_id}/results — metrics overview.
3. GET /tests/{test_id}/recommendations?limit=0 — full recommendation export.
4. POST /tests/{test_id}/report/generate-reportlab → poll GET …/report/status → GET …/report/download — PDF report.

Python SDK

The SDK does not yet ship typed helpers for these three read routes. Use curl or httpx with the same REDRAVEN_API_KEY, REDRAVEN_ORGANIZATION_ID, and REDRAVEN_BASE_URL as in the Python SDK guide.