Hello Tomorrow Deep Tech Pioneer 2026

The FICO score
for AI-generated code

Calibrated 0–100 reliability scoring backed by empirical outcome data. Track what happens to AI code after it ships.

4,700+
Tracked commits
350+
Repositories
19
Languages
v2.3
Model version

Three capabilities

🔍

Detection

Identify AI-generated commits from Claude, Copilot, ChatGPT, Cursor, Devin, Aider, Windsurf, Gemini, and more. Commit signatures, author patterns, code markers.

📊

Scoring

Calibrated reliability scores from real outcome data—CI pass/fail, reverts, fix commits. Wilson confidence intervals, ECE/Brier metrics. Not opinions, empirics.

🛡

Audit

Complete audit trail for EU AI Act compliance (Article 50, Aug 2026). Which tool generated the code, when, what score, what happened after deployment.

Install in 60 seconds

# .github/workflows/ship-score.yml # Copy one file. Scores every PR automatically. name: SHIP Score on: pull_request: types: [opened, synchronize, reopened] # Full file: # github.com/vibeatlas/.../ship-score.yml
# Score any repo from your terminal $ npx ship-score facebook/react SHIP Score Report Score: 40/100 (F) Confidence: 57% Risk Level: HIGH Sample Size: 30 commits $ npx ship-score rust-lang/rust SHIP Score Report Score: 89/100 (B) Confidence: 70% Risk Level: LOW Sample Size: 27 commits
# Works with Claude Code, VS Code, Cursor, any MCP editor $ npx -y @vibeatlas/ship-mcp-server # Or add to your MCP config: { "mcpServers": { "ship": { "command": "npx", "args": ["-y", "@vibeatlas/ship-mcp-server"] } } }