Evidence-Based, Independent Evaluation
SpecGrade evaluates spec-driven AI coding tools using systems engineering criteria including requirements quality (INCOSE, EARS), traceability (RTM), verification gates (V&V), change control, and portability.
Every score is backed by evidence—documentation, repository files, or test runs. All evaluations are independent; sponsors never influence scores.
We track 15+ tools including GitHub Spec Kit, Kiro, OpenSpec, BMAD-METHOD, Tessl, Traycer, and more. Each receives a detailed scorecard with rationale and citations.
What We Evaluate
Requirements Quality
From free-text prompts to templated, testable acceptance criteria (EARS-like)
Traceability
Bidirectional requirement↔task↔code↔test links kept current
Verification Gates
Enforced gates where tests from acceptance criteria must pass
Change Control
Delta proposals with review and audit logs versus overwriting specs
Portability
Works with any agent, open formats, MIT/Apache licensing
Maturity
Active development, releases, documentation, and community
Weekly Newsletter
Get tool updates, new scorecards, and insights on spec-driven development. No spam, evidence-based analysis only.
Directory and scorecards coming soon. Stay tuned for detailed evaluations of 15+ spec-driven AI coding tools.