The first tool that turns every line of AI-generated code into a verifiable, auditable, and certifiable product — from the very first prompt, in your CLI or directly inside your coding agent.
Built for healthcare, finance, and regulated software development
❌
Hospitals, banks, and government agencies hesitate to buy AI-generated software because there's no trusted verification standard.
❌
Most developers hide the fact they used Claude, Cursor, or Grok — slowing adoption and breeding distrust.
❌
Traditional audits happen too late and are expensive. Compliance teams want proof from line one.
TraceCert runs alongside your AI coding agent from the first line of code. Every change is continuously audited, logged, and certified — turning the build process itself into a verifiable compliance trail.
$ tracecert scan . --level 4 --export-pdf Scanning project... Generated trace -- Running certification... ================================================================ TraceCert Level 4 Verification -- Manifest Signed ================================================================ Verification Level : 4 Levels Verified : [1, 2, 3, 4] AI-Generated : 99.0% (631 lines) Verification Score : 91/100 Secrets Found : 0 HIPAA Score : 78/100 [PHI risk: MEDIUM] HIPAA-ready : True SOC2 TSC Alignment : 82/100 [A] AI Quality Grade : B (slop score 74) Signature Type : Ed25519 Output File : .tracecert/cert-manifest.json PDF Report : .tracecert/report.pdf ================================================================
AI-Built
Trace parsed, AI provenance recorded, basic quality checks and disclosure.
Security Verified
Secret detection, OWASP no-secrets check, full provenance chain.
Domain Grade
HIPAA risk scoring, SOC 2 TSC alignment, AI quality (slop) detection.
Most PopularEnterprise Certified
Combined SOC 2 TSC + HIPAA signal in a single signed attestation. Requires Level 3 thresholds met.
Now AvailableTraceCert now ships as an MCP server, so Claude Code, Claude Desktop, Cursor, or any MCP-speaking agent can scan and certify code as native tool calls — no context-switching to a terminal, no copy-pasting output back in.
scan_file, scan_diff, and get_policy run deterministic, no-LLM checks in under a second — safe for an agent to call after every single change.
full_certify runs the complete Level 1–4 pipeline server-side and returns a job id to poll — call it at task completion, not mid-loop.
# Claude Code — add the server
claude mcp add --transport http tracecert \ https://mcp.tracecert.ai/mcp \ --header "Authorization: Bearer $TRACECERT_API_KEY"
# Bundle & certify a snapshot
git ls-files --cached --others --exclude-standard -z \ | tar --null -T - -czf upload.tar.gz
Works With
Keys are issued directly — no self-serve signup yet.
Define minimum scores, secret gates, and custom rules in .tracecert/policy.yaml. Block CI merges automatically when thresholds aren't met.
Automated GitHub PR comments with compliance scores and commit status checks. Run tracecert pr-check in CI to gate every merge.
Export audit-ready PDF reports and certification badges for procurement, investor diligence, or compliance reviews with --export-pdf.
Real PHI risk phrase detection, HIPAA §164.312 technical safeguard mapping, and compliance scoring out of the box.
Maps your codebase against SOC 2 Trust Services Criteria (CC6, CC7, CC8, A1, C1, PI1, P1) and produces a graded alignment score.
Ed25519-signed, tamper-evident manifests with model name, prompt hash, content hash, and git revision — a full chain of custody for every AI-built artifact.
Native tool calls for Claude Code, Claude Desktop, and Cursor — scan_diff and full_certify without leaving your agent. See how it works →
Free to use. Powerful enough for regulated environments.
# Install from the wheel in this repo
$ pip install dist/tracecert-0.5.0-py3-none-any.whl
# With dashboard + audit pack support
$ pip install "dist/tracecert-0.5.0-py3-none-any.whl[enterprise]"
# Run your first certification
$ tracecert scan .
TraceCert is proprietary software. The full tool — all features, all levels — is available for evaluation. Commercial use requires a license. No tiers, no feature gates, no hidden limits.
Evaluation
Free
Personal use & internal evaluation only
Commercial License
Contact Us
Production use • Teams • Regulated environments