agenticevals.dev

Choose your pack

Four products. One independent answer.

Agent-powered testing, directed and signed off by Leo Ruocco, with 30 years' experience building controls across financial services, insurance and defence. Choose the Honest Second Opinion for evidence you already hold, the Deny Means Deny certificate for one refusal boundary, the advanced pack for deeper tests on open-source or private code, or a deep evaluation scoped to your product. Pay securely with Stripe, or enquire by email for a deep evaluation, then reply with the commit you want tested (or say 'latest main'). We read the code and send you a short approach specification to agree before testing.

Four products

The Honest Second Opinion reviews evidence you already have. The Deny Means Deny certificate is one focused refusal-boundary evaluation. The advanced pack keeps the five-check baseline and adds deeper tests, built for your product. The deep evaluation is scoped per customer and starts with an email. Delivery is confirmed in your approach specification.

No VAT is added. The price shown is the total you pay.

Refunds follow our Founders Pragmatic Refund Policy.

Independent review · Honest Second Opinion
£299
An honest, independent second opinion on the evidence you already have: what holds, what does not, and what we would test next. Written and signed by Leo Ruocco. This is a signed written opinion, with no certificate.

Scope: one product as it stands today, one repository or one live app.

Open source or closed source?

Closed source, until 14 November 2026. Early-bird price until 14 November: £299. From 14 November, closed code is £599 (NDA, no-training AI and source deletion). Buy with the same link as the button above.

Buy the Honest Second Opinion · £299 See an example
Certificate · Deny Means Deny £399 An independent check that “denied” really means denied at one fixed commit, with a signed Deny Means Deny certificate when the check holds.

Scope: one refusal boundary at one fixed commit.

Open source or closed source?

Closed source, until 14 November 2026. Early-bird price until 14 November: £399. From 14 November, closed code is £799 (NDA, no-training AI and source deletion). Buy with the same link as the button above.

Private-hosted evaluation available on request: we run your evaluation on a dedicated machine rented for your job alone, using open models, so your code is never sent to an AI provider, and the machine is wiped afterwards. Ask us: sales@agenticthinking.uk.

Get the certificate · £399 See an example
Open source · advanced pack£499Everything in the five-check baseline, plus the deeper tests: the full record chain link by link, advanced methods, repository-specific cases built for your requirements, negative controls, findings with remediation, and a complete evidence index.

Scope: one agreed workflow or control boundary at one fixed commit, set out in the approach specification you agree before testing.

Buy the advanced pack · £499See an example
Private code · advanced pack£999Everything in the five-check baseline, plus the deeper tests, under our private-code terms: NDA before access, only no-training AI, and source-code deletion confirmation under our published policy.

Scope: one agreed workflow or control boundary at one fixed commit, set out in the approach specification you agree before testing.

Private-hosted evaluation available on request: we run your evaluation on a dedicated machine rented for your job alone, using open models, so your code is never sent to an AI provider, and the machine is wiped afterwards. Ask us: sales@agenticthinking.uk.

Buy the private advanced pack · £999See an example
Deep evaluation · scoped per customer from £2,500 A deeper evaluation scoped with you before any work starts: the concerns you actually have, the boundary you need tested and the evidence you need to see. Scope and price are agreed in writing first.

Scope: agreed with you in writing; quoted per customer.

Email us about a deep evaluation See an example

Enquiry by email only. No buy button: we scope and quote each deep evaluation per customer. The advanced pack example shows the report format a deep evaluation extends.

Fixed prices for fixed scopes. If your job is bigger, we say so before any extra work starts and quote the extra as a deep evaluation. You never pay more for the product you bought.

The Standard pack has been replaced

If you followed an old Standard link, that product is no longer on this buy menu: it has been replaced by the Deny Means Deny certificate. Anyone already quoted a Standard price keeps it: the old Standard links still work (open source or private code), or email us.

Which pack? Answer four questions to see which product fits. When you want a paid evaluation, choose a pack above.

See what a baseline report contains. The advanced pack example shows a fictional report format; actual coverage follows the purchased package.
Need more than the advanced pack, or not sure which? Get started. No sales calls.

Agentic Evals independently evaluated seal

Included evidence

A signed outcome, not another self-assessment.

An advanced pack or deep evaluation receives a signed Evaluation Record. When every applicable check passes, we also issue a Certificate of Independent Evaluation tied to the tested commit. The Deny Means Deny check issues its own Deny Means Deny certificate when the check holds. The Honest Second Opinion is a signed written opinion, with no certificate. Use the certificate in an enterprise review, investment discussion, trust centre or release pack.

The certificate is a scoped, point-in-time record. It is not accredited product certification or a general safety guarantee.

View the certificate specimen →

The advanced pack

Everything in the five-check baseline. Advanced tests. Your specific boundary.

One package, all three layers: the five-check baseline is the foundation, not an extra purchase. The deeper tests are built specifically for your requirements: your approach specification defines them. Directed and signed off by Leo Ruocco, with 30 years' experience building controls across financial services, insurance and defence.

The published case inventory, a 10-case reference baseline and a 66-template advanced catalogue in 12 families, is on how we evaluate.

Code-defined cases exercise your software. The advanced pack retains the five baseline checks and adds selected advanced methods and agreed repository-specific cases. Your software's records are compared with independent observations, producing a report and reproducible evidence.
Plate IV. The five-check baseline is retained in the advanced pack, with selected advanced methods and agreed repository-specific cases added. Target records are compared with independent observations. Named cases are source-labelled reference examples, not customer results or guaranteed coverage.
01 · Five-check baseline

Baseline included

The five-check baseline framework, commit and environment record, reviewed outcomes and supporting evidence. Target-class exclusions are explained.

02 · Advanced methods

Relevant deeper tests

Applicable bypass, approval-substitution, replay, crash/retry or evidence-integrity tests, plus positive controls for authorised behaviour. Not every test family applies to every product.

03 · Repository-specific cases

Built for your product

The code, declared controls and configuration shape the agreed cases for one workflow or control boundary at a fixed commit.

You receive: a reviewed PDF and HTML report; pinned target-specific test cases, configuration and reproduction instructions; supporting raw evidence with an index; founder-actionable remediation guidance and explicit limits. Sensitive evidence is handled under the agreed disclosure terms. The test handover covers the agreed scope, not our entire proprietary rig.

Agree the scope first: buy the pack, then reply with the commit you want tested (or say 'latest main'). We read the code and send you a short approach specification to agree before testing. You implement fixes. This is not an unlimited audit or a runtime-development service.

Scope and delivery date are confirmed in your approach specification. If you are unsure your product fits, ask about package fit before buying. These evaluations are not implementation services or penetration tests. Any certificate records the scoped evaluation outcome at one commit; it is not accredited product certification. Buying a pack means you accept our terms.