Stop Shipping Bugs That Hide in Your Spec

ReqProof mathematically proves your requirements are consistent, complete, and realizable before you write a single line of code. Found a crash bug in Tyk API Gateway that 400 lines of code review missed.

NASA FRET Heritage DO-178C Ready Git-Native
reqproof verify
$ reqproof verify

Your Requirements Are a Liability,
Not an Asset

For safety-critical engineering teams dissatisfied with manual requirements processes, these three pain points are unavoidable, urgent, and dangerously underserved.

Specifications Rot in Silence

Requirements live in Confluence. Code lives in Git. Tests live in CI. Nothing connects them. When code changes, nobody updates the spec. When the spec changes, nobody updates the tests. Drift is invisible until it causes an incident.

The Hardest Bugs Live in the Spec

Most testing catches implementation bugs. But what about requirements that contradict each other? Or edge cases nobody specified? The costliest defects are the ones you never wrote a test for because the spec never asked the question.

Certification Is Manual Pain

DO-178C, ISO 26262, and IEC 62304 demand traceability matrices, MC/DC evidence, and SRS documents. Teams spend months assembling paperwork that is outdated the moment it is printed.

The Verification Chain

One command. Six verification layers. Each catches a different class of defect that no amount of testing can find.

1

Write Requirements

Define requirements in YAML with FRETish temporal logic. Or write plain English and let AI translate with 86% accuracy.

2

Prove Correctness

Kind2 and Z3 solvers prove realizability, consistency, and data property correctness for all possible inputs. Not testing -- mathematical proof.

3

Find the Gaps

Gap analysis flags every unconstrained output. It forces the questions nobody thought to ask -- like the one that found the Tyk crash bug.

4

Generate Tests

FLIP generates MC/DC test traces. Z3 produces boundary-value fixtures from solver models. Automatic, exhaustive, from formal proofs.

5

Link to Code

Comment-based annotations in 20+ languages. Automatic bidirectional traceability. Suspect link detection when things drift.

6

Ship with Proof

Generate NPR 7150-compliant SRS documents with traceability matrices, coverage heat maps, and verification evidence.

verification output
$ reqproof verify ReqProof Verification Summary ------------------------------ Requirements: 551 total Step Status Details ---- ------ ------- Validate PASS 551/551 valid Realizability PASS all components realizable Consistency PASS 0 contradictions Vacuity PASS 0 vacuous Gap Analysis WARN 9 unconstrained outputs Lint PASS 0 untraced functions Suspect Links PASS 0 stale links Overall: PASS

Everything You Need to Verify Requirements

A complete formal verification toolkit that fits into your existing developer workflow.

Formal Verification

Prove requirements are realizable, consistent, and complete using Kind2 and Z3 SMT solvers. Not testing -- mathematical proof that catches contradictions no test suite can find.

Bidirectional Traceability

Every requirement links to code and tests via simple comments. // SYS-REQ-001 in any language. Automatic linking. Suspect link detection when things drift.

AI-Assisted Authoring

Write requirements in plain English. ReqProof translates to formal FRETish notation with 86% accuracy, validates instantly, and suggests fixes. No formal methods expertise required.

Development Workflow Engine

reqproof verify runs the entire pipeline in one command. reqproof ci for GitHub Actions and GitLab CI. Fits your existing developer workflow.

Multi-Language Support

Comment-based traceability scanning in 20+ programming languages. Go, Python, Java, C, C++, Rust, TypeScript, and more. Works with any codebase.

Certification-Ready Documents

Generate NPR 7150-compliant SRS documents with traceability matrices, coverage heat maps, and verification evidence. Maps to DO-178C Tables A-1 through A-10.

Applied to Tyk API Gateway

A real production system. 400 lines of Go. 2 hours of specification work. Here is what ReqProof found.

0
Crash bug found
(nil pointer dereference)
0
Undocumented decisions
made explicit
0
Formal requirements
verified mathematically
0
Of specification work
to find the bug

"The bug was found because gap analysis forced the question: 'what happens when the store is unavailable?' Nobody had asked until the spec required an answer. The code never checked for nil -- and it had been in production for years."

-- ReqProof Case Study: Tyk API Gateway Policy Engine

Certification-Ready from Day One

Built on NASA FRET formal methods heritage. Generates compliance artifacts for the standards that matter.

DO-178C
Aviation Software
ISO 26262
Automotive Safety
IEC 62304
Medical Devices
NPR 7150.2D
NASA Software

Section-by-section mapping for DO-178C Tables A-1 through A-10, ISO 26262 Part 6/8, IEC 62304, and 7 additional standards. SRS documents include coverage heat maps, traceability matrices, and verification evidence.

ReqProof vs IBM DOORS

A fundamentally different approach. Not a cheaper copy -- a new category of tool that DOORS cannot replicate.

Capability IBM DOORS ReqProof
Storage Proprietary database Git (YAML files)
Formal verification None Kind2, Z3, FLIP
AI assistance Watson quality checks NL-to-FRETish (86% accuracy)
Price $200-300/user/year Open core
Setup time Weeks + consultants reqproof init
Code traceability Manual Auto-scan (20+ languages)
CI/CD integration Limited reqproof ci
Test generation None MC/DC + Z3 boundary fixtures
SRS generation Manual export Auto (NPR 7150, heat maps)

Stop Guessing. Start Proving.

Join engineering teams that verify requirements with mathematical proof, not hope.