The SAT scorefor your codebase

Sprinno tells you, in one number, how your codebase actually stacks up — and shows you exactly what's dragging your score down. Think of it as the SAT score for your codebase.

What we measure is public: the full list of metrics and how each one is detected is in our open quality ontology. What's not public is how those metrics are weighted into your final score — that's calibrated from data across hundreds of repos and kept private, the same way a credit score's inputs are public but the formula isn't. Publish the weights and the score stops meaning anything — teams start optimizing for the formula instead of for actually shipping better code.

Add the Sprinno MCP server (https://api.sprinno.xyz/mcp) to GitHub Copilot in VS Code, Copilot CLI, the GitHub Copilot app, Lovable, v0, Replit, Claude Code, Cursor or Codex, and sign in once. GitHub-hosted Copilot coding agent does not yet support the OAuth sign-in Sprinno requires. Setup instructions for each supported platform are at sprinno.dev/docs/mcp.

We use Latent Capability Scoring (LCS) to combine open, publicly-documented metrics — test coverage, dependency freshness, architecture signals, and more — into one score. The metrics themselves aren't a secret; you can see all of them. What's proprietary is how heavily each one is weighted, calibrated empirically from real repo data and kept private for the same reason above. Your score is also contextual to your stage — what's good for a seed startup differs from a Series B company.

Your agent uploads the project over an encrypted connection. It is unpacked in an isolated workspace on our servers, scanned by a deterministic engine, and deleted as soon as the scan finishes, whether it succeeds or fails. We keep scores, file hashes and the file paths and line numbers behind each finding, never your source code. No AI model sees your code during scoring.

Developers and engineering teams who want an honest picture of their codebase quality. Run it to find your weak spots, track improvement over time, and benchmark against peers. Investors and due-diligence firms also use Sprinno Cloud for objective assessments during fundraises and M&A.

Static analysis tools check code against fixed rule thresholds — they're deterministic but shallow. Sprinno goes far beyond code-level checks: we assess process maturity, architecture quality, resilience, and documentation. Our weights are empirically derived from data across hundreds of repos, not arbitrary thresholds set by experts.

When your agent uploads a project, you get a private score you can track against your own history. Link the project's GitHub repo — including the repo Lovable, v0 or Replit syncs to — and Sprinno scans that repo itself on every push. That scan becomes the verified score of record, unlocks history-based signals, and agent uploads become quick previews.

Not easily. Your agent only sends files; every signal is computed on our servers, so there is nothing to edit on the way. Files that don't match the declared hashes are rejected, inline suppressions don't change the score, and a project that is swapped for different code starts a new history line instead of showing as an improvement. The weights that combine metrics into a score are private, so there is no single metric to juice.

Metrics are stack-agnostic at the pillar level. Whether you write Python, TypeScript, Go, Rust, Java, or anything else, we measure the same underlying engineering behaviors. Stack-specific calibrations exist only for language-level metrics where normalization is needed.

Company Score is a weighted, multi-repo quality score. Assign each repo a role — Core Backend, Core Frontend, CLI/SDK, Supporting Service, etc. — and Sprinno computes a single company-wide score with per-bucket breakdowns. It accounts for your funding stage and the relative importance of each repo to your business. Available on Pro ($15/member/mo, 4 members minimum).

Yes. Every finding comes with the file and line behind it and how to fix it, and your agent can read them through the same connection. Ask it to fix the top findings, then score again: Sprinno compares the two versions and only calls a change an improvement when it is larger than the measurement error.

After two or more scans on the same repo, Sprinno computes a graph delta — showing which components were added, removed, or changed, which findings were resolved or introduced, and how the score moved. It's the diff for your architecture, not just your code.

Sprinno badges are live-updating SVG images that show your quality score. Embed them in your README, landing page, or pitch deck with a single HTML snippet or Markdown line. Badges are available on Pulse plans and above, and include a verification link back to your Sprinno profile.

Yes. Pro users can export scans as self-contained HTML reports, generate investor PDFs, and create unlimited share links. Pulse includes 10 share links. The canvas supports four visual presets (Diligence, Investor, Signal Flow, Blueprint), and you can walk through guided chapters for presenting to non-technical stakeholders.