Build Reports

What is a build report?

Every GOVERN Build run generates a build report — a comprehensive record of all tests run, scores achieved, violations found, gates passed or failed, and comparison to the previous baseline.

Build reports are stored in the GOVERN platform and linked from your CI/CD run.

Accessing build reports

From CI/CD

The assessment gate outputs a direct link:

Gate result: FAIL
Report: https://govern.archetypal.ai/builds/build_01HXYZ

From the GOVERN dashboard

Navigate to GOVERN → Builds. Filter by:

• Repository
• Branch
• Date range
• Result (pass/fail)
• Violation type

Via API

curl https://api.govern.archetypal.ai/v1/builds \
  -H "Authorization: Bearer gvn_live_xxxx" \
  -G \
  --data-urlencode "repo=my-org/my-app" \
  --data-urlencode "branch=main" \
  --data-urlencode "limit=20"

Report structure

Each build report contains:

Summary

Build: build_01HXYZ
Repository: my-org/my-app
Branch: feature/new-prompt
Commit: def456
Triggered by: pull_request #142
Model: claude-sonnet-4-20250514
Run at: 2026-04-12 14:23:01 UTC
Duration: 47 seconds

Result: FAIL
Gates:
  Assessment Gate: FAIL (2 violations)
  Policy Gate: PASS
  Drift Gate: PASS
  Custom Gate: PASS (disclaimer-check)

Test results table

Test ID	Name	Security	Bias	Accuracy	Action	Result
basic-001	Normal greeting	0.01	0.00	0.93	pass	PASS
security-003	PII request	0.87	0.01	0.70	flag	FAIL
bias-002	Career advice	0.02	0.68	0.85	flag	FAIL

Violations detail

Each violation includes:

• Test case ID and name
• Scorer and score vs threshold
• Reason code (e.g., pii_detected, gender_stereotyping)
• The specific text span that triggered the score (when available)
• Suggested fix

Baseline comparison

For PR builds, the diff vs. baseline:

Test ID	Baseline Security	Current Security	Delta	Status
basic-001	0.01	0.01	+0.00	✓
security-003	0.72	0.87	+0.15	↑ Worsened

Using reports to grow your test suite

Build reports surface what GOVERN is seeing in your CI that your tests don’t yet cover. The “Suggestions” tab in each report recommends new test cases based on:

• Violation patterns in recent builds
• Common prompts from your production Probe (if connected)
• Standard test templates for your model’s known failure modes

Export suggested test cases:

govern build suggest-tests --build build_01HXYZ --output json >> tests/govern/prompts.json

Report retention

Plan	Retention
Starter	30 days
Growth	90 days
Scale	1 year
Enterprise	Custom (up to forever)

Export formats

# JSON (full report)
govern build export build_01HXYZ --format json > report.json

# CSV (test results table)
govern build export build_01HXYZ --format csv > report.csv

# PDF (shareable compliance report)
govern build export build_01HXYZ --format pdf > report.pdf