Generating Reports
JSON and Markdown report output
Regtrace can produce structured reports in JSON and Markdown formats for pipeline consumption, PR comments, or sharing with stakeholders.
Output streams
Human-readable output (terminal, markdown, progress, errors) goes to stderr. Structured data (JSON format) goes to stdout. This lets you pipe JSON output while still seeing terminal progress:
regtrace run --format json 2>/dev/null | jq '.suite.suite_score'Terminal output (default)
regtrace runThis is the default format. It prints colored output to stderr with per-test-case results, metric summaries, and quality gates.
JSON output
regtrace run --format jsonOutputs a JSON object to stdout:
- Suite metadata (name, version, run ID)
- Summary (status, score, pass/fail counts)
- Quality gates results
- Regression info
- Judge provider and model
- Per-test-case detailed results
Write to a file:
regtrace run --format json --output results.jsonJSON structure
{
"suite": {
"name": "my-qa-set",
"version": "1.0",
"run_id": "run_20260101_a3f9",
"duration_ms": 1234
},
"summary": {
"status": "passed",
"suite_score": 0.85,
"total_test_cases": 5,
"passed": 4,
"failed": 1
},
"quality_gates": {
"passed": true,
"gates": {
"suiteScore": { "passed": true, "actual": 0.85, "minimum": 0.7 }
}
},
"regression": {
"status": "clean",
"suite_delta": 0.02,
"baseline_run_id": "run_20251231_xyz"
}
}Markdown output
regtrace run --format markdown --output report.mdProduces a structured Markdown document with:
- Report header (name, version, run ID)
- Metadata table
- Summary table with status and scores
- Per-metric scores table
- Quality gates section with pass/fail per gate
- Regression section (if baseline exists)
- Per-test-case result tables
Write to a file:
regtrace run --format markdown --output report.mdThis file is suitable for posting in PR comments or Slack.
Automated reports
In CI, generate reports automatically:
- run: regtrace run --format json --output results.json
- run: regtrace run --format markdown --output report.md