Deterministic Verification Engine
Deterministic test verification for AI-generated codebases.
AI coding tools create test suites that pass continuous integration by asserting shallow existence checks (.toBeDefined(), .toBeTruthy()) and skipping critical domain error branches. The build stays green while regressions escape.
Pramāṇa audits test suites using deterministic AST analysis and synthetic fault injection. It proves surviving mutants and delivers an engineer-ready assertion hardening git patch within 72 hours. Audits run asynchronously without sales calls or repository write access.
72-hour turnaround · Asynchronous delivery · Read-only access
The delivery loop
Four verified deliverables: Static Assertion Ledger, Error Branch Matrix, Fault Resistance Proof, and Assertion Hardening Git Patch.
How the audit deliversEmpirical Evidence
Verified test gaps in public open-source repositories.
Each exhibit below documents physical execution against public test suites. When logic was mutated, native test suites exited with code 0, confirming that broken logic passed green in CI.
Security specification · Webhook signature verification
standard-webhooks/standard-webhooks
30/100: 7 of 7 domain security error branches unasserted
All negative tests assert only generic WebhookVerificationError. Signature bypasses and expired timestamps pass CI under the same green assertion.
E-commerce platform · Cart & order calculation
medusajs/medusa
Surviving Mutant: Tax calculation reversed (MathBN.add → MathBN.sub)
Mutating tax addition to subtraction in order calculation passed the full test suite with 100% green status. A financial reversal escaped CI.
Communications API · Batch email transmission
resend/resend-node
13 Vacuous Matchers: 13 instances of .toBeDefined() masking payload validation
Core batch sending tests assert only that network calls and headers 'exist'. Malformed or empty payloads continue to pass green.
Commercial offers
Flagship audit
Pramāṇa AI Test Suite Audit
A deterministic audit of your AI-generated test suite. We uncover unasserted error branches, eliminate vacuous matchers (.toBeDefined()), prove surviving mutants, and deliver an engineer-ready assertion hardening git patch.
£2,500 fixed fee · 3 working days from complete evidence
See audit deliverablesContinuous protection
Pramāṇa Continuous CI Gate
Automated pre-merge verification that blocks unasserted error branches, vacuous matchers, and surviving synthetic mutants on pull requests.
£499/month · Continuous pre-merge GitHub Action gate
See CI action gateAutomation
Automation & AI Ingestion
Controlled AI automation across documents, messages, forms, spreadsheets, APIs and system events, built in the client's Google Cloud project with validation gates, a named exception owner and an independent action record.
See the automation serviceAutomation use case
Controlled procurement workflows
Automation prepares request intake, quote comparison, written checks and approval routing. Named people approve, and the workflow takes only the purchase actions written into its specification.
Founder Historical Track Record
Outcomes measured on earlier engagements.
Outcomes recorded in engagement closeout reports (2014–2024). These historical records demonstrate founder domain authority and why Pramāṇa was built, not active consultancy services.
Telecom · Backend engineering
30%
Monthly compute cost reduction
Erlang backend rewrite
Measured outcome · anonymised client · recorded in the engagement closeout report
Read the engagement recordHealthcare · Quality engineering
3×
Faster UAT cycle (15 days → 5)
Patient portal automation
Measured outcome · anonymised client · recorded in the engagement closeout report
Read the engagement recordFounder Provenance
Ten years of release engineering and platform delivery.
Pramāṇa was built after ten years leading quality engineering and release architectures across telecom, regulated healthcare, and high-scale platforms. A selected retail programme covers Next.js migration, GCP infrastructure, microfrontends and CI/CD modernisation.
Selected records include work delivered before Anystack Engineering was incorporated in May 2024. Each record states its provenance; these results are not evidence of current commercial models.
Engineering research
Deterministic verification methodology.
We apply peer-reviewed mutation testing research and formal verification standards across product development, quality engineering, and platform reliability.
Review the sourcesCompany assurance