> Governance rails for AI-agent development: no self-attestation, receipts required for every completion claim, and fail-closed handling of unverifiable output. - Canonical: https://hadalinstruments.com/instruments/agent-governance-constitution/ --- [Back to Instruments](https://hadalinstruments.com/instruments/) P5 The Honesty Stack # Agent-Governance Constitution The rails that caught a forged ratification — packaged. Anticipated price Anticipated · not ratified · not an offer From £39/seat/mo — org licences from £6,000/yr - **Annual term** — from £390/seat/yr Twelve months of service for the price of ten. - **Billed monthly** — from £468/seat/yr from £39/seat/mo × 12. Monthly billing costs twenty per cent more than the annual term — the uplift is on monthly, it is not a discount on annual. - **The difference** — from £78/seat/yr Two months in twelve. Paying annually saves two months, not twenty per cent — the two are different numbers and only one of them is true. Open to academic access AI-agent governance is an active research area, and my own founding incident is a first-party case study in it. Researchers studying agent oversight should not be paying full price to examine the constitution that caught a fabricating agent. Anticipated academic rate: 50% of the annual list price — from £195/seat/yr . The academic rate applies instead of the seat and term bands, never on top of them. Stacked, the three would land below the cost of serving the account, and a rate I cannot honour is worse than one I never offered. [Academic access — who qualifies and on what terms](https://hadalinstruments.com/academic/) Nothing on this site is on sale. There is no checkout, no cart and no payment link on any page. Prices publish pre-launch so they can be read, compared and checked rather than requested — [seat bands, group licences and multi-year terms are set out in full below](#commercial). here would have added one landmark per placement, which on the homepage means a landmark list that is mostly disclaimer. --> Standing policy — read this beside the figure - **No promise of profit, ever.** Hadal makes no performance claims and carries no implied edge. Past measurements describe instrument behaviour — never future returns. - **You own risk management.** Hadal cannot control it and does not insure it. Good tools do not fix bad discipline — and this site says so. - **Analytical tools, for discretionary use.** Nothing here is investment advice or a recommendation to trade. Every decision, and every outcome, is yours. The same statement stands in the footer of every page.[Terms](https://hadalinstruments.com/terms/) [Privacy](https://hadalinstruments.com/privacy/) [How it measuresThe Honesty Stack suite methodology](https://hadalinstruments.com/methodology/stack/)[BoundsWhat it does not establish](#what-it-does-not-do) Published measurements NOT YET PUBLISHED Specification | Value | Catalogue no. | P5 | Suite | [The Honesty Stack](https://hadalinstruments.com/suites/stack/) | Methodology | [How the Honesty Stack suite measures](https://hadalinstruments.com/methodology/stack/) | Availability | Pre-launch — not on sale | Anticipated price | [from £39/seat/mo · from £390/seat/yr](#anticipated-price) | Delivery | Seat licence · deployment licence | Published measurements | NOT YET PUBLISHED | Provenance artifact | NOT YET PUBLISHED | In July 2026 an AI agent working on my core engine forged operator sign-offs, fabricated statistical results, and declared the system “COMPLETE” when nothing worked. I caught it — and the thing that caught it is this product. The Agent-Governance Constitution is the discipline stack I now run on every agent session, packaged for teams who build with AI agents and cannot afford to discover this failure mode in production. It is not a prompt. It is a set of rails that assume the agent will optimize for “task complete” over “task true”, and make that optimization visible and fail-closed. ## What ships The [Burned Door](https://hadalinstruments.com/glossary/burned-door/) incident decided which rails ship and why each one has to be structural. - **The constitution itself.** The binding disciplines: never fabricate, never self-attest completion, never mint an approval that belongs to a human operator. Adapted to your codebase, with protected paths and operator-act boundaries defined. - **The receipt rule.** A deliverable counts as done only with a verifiable receipt — a passing build, an observed behaviour, a content-hashed artifact. “The agent said it works” is not a state your pipeline can reach. - **Fail-closed enforcement.** Hard rails in CI, not soft rules in a prompt file: gates that fail the build on fabrication markers, unverifiable provenance, and self-attestation language. Enforcement lives where the agent cannot bypass it. - **The post-mortem.** The [Burned Door incident](https://hadalinstruments.com/research/the-burned-door/) that produced this system, published as first-party testimony with its forensic evidence pack still NOT YET PUBLISHED — because a governance product whose origin story is hidden would be its own counterexample, and one that overstated its own receipts would be a different counterexample. ## What it does not do The Agent-Governance Constitution does not make agents honest. Nothing does. It makes dishonesty expensive, visible, and non-fatal: an agent that fakes a gate fails a build instead of shipping a lie. ## Who it is for Teams shipping with coding agents, quant shops automating research pipelines, and anyone whose CI currently believes whatever the agent tells it. ## The ship gate No instrument is sold until it does what this page says it does. Where a page is written in the future tense, that tense is a statement about timing rather than a hedge about capability: the instrument is not finished, so it is not listed as available, not priced as available, and not sold. It waits. Nothing described in this catalogue is a placeholder that will quietly disappear. An instrument that turns out to be wrong gets a [kill-ledger entry](https://hadalinstruments.com/kill-ledger/), not a deletion — which is the only version of that promise anyone can check. ## Commercial terms Published in full, pre-launch, so they can be read and checked rather than requested. Every figure is an anticipated indication I have set and not yet ratified, and every derived figure is the arithmetic of the one above it — shown, not asserted. Nothing here is purchasable: there is no checkout on this site. ### Why Agent-Governance Constitution is priced the way it is - **What the figure buys** — The rails, adapted to your codebase rather than handed over off the shelf: the constitution itself with your protected paths and operator-act boundaries defined; the receipt rule, under which a deliverable counts as done only against a verifiable receipt — a passing build, an observed behaviour, a content-hashed artifact; and fail-closed gates in CI that break the build on fabrication markers, unverifiable provenance and self-attestation language, installed where the agent cannot reach them. It also includes the published post-mortem of the incident that produced the system. The forensic evidence pack behind that post-mortem is not yet published, which is stated here rather than left to be discovered, because a governance product that overstated its own receipts would be its own counterexample. What is not included is agent honesty. Nothing sells that; the rails change what an agent can ship, not what an agent will attempt. - **Why it is priced this way** — Per seat at the entry tier because a seat is a person running agent sessions, and each of those people is a separate stream of completion claims arriving at the gates. Org licences exist because past a certain size the thing being licensed stops being individual practice and becomes a property of the estate — one CI configuration, one set of protected paths, a rule the whole pipeline obeys — and an additional engineer joining a rule that already exists costs Hadal nothing to serve, so billing for them would be billing for nothing. The org tier runs annually because a governance rule renegotiated every month is not really a rule. - **What the alternative costs** — Write it yourself. That is entirely possible, and this page states the disciplines rather than hiding them; a team that adds its own receipt rule and its own CI gates gets a real distance down the road. What is being sold is the incident that produced the list — which rails matter, and why each one has to be structural: that self-attestation must be unavailable rather than discouraged, that enforcement must live where the agent cannot edit it, that a missing receipt must fail closed rather than warn. The other alternative is the current default, a pipeline that believes whatever the agent tells it, which costs nothing right up until the first completion claim that was not true, and then costs whatever shipped behind it. Every figure on this page is an anticipated indication awaiting ratification, and nothing here is purchasable. The reasoning above is published for the same reason the arithmetic below is: a price you can interrogate is worth more than a price you have to accept. ### The two-SKU split One route. This instrument has no marketplace equivalent, and the cell says so rather than sitting empty — an empty cell reads as an omission, and this is a fact about how the instrument is sold. Route | What it is | Anticipated | Marketplace SKU | None. This instrument is licensed directly, per seat or per deployment against the systems it runs on, and no platform marketplace sells that shape. There is no third-party SKU between you and it. | NO MARKETPLACE ROUTE | Licence (this site) | Licensed directly from the hub — per seat where the instrument is used by people, per deployment where it is installed against systems. Disclosed renewal terms, cancellable in one step, carrying the receipt trail the instrument produces. | from £39/seat/mo · from £390/seat/yr | The price rule The at-or-below-marketplace cap binds the instruments that ship through a platform. This one does not, so the cap has nothing to bind. What holds in its place is simpler and checkable: the figure above is the only figure. There is no separate, higher list price it is discounted from, nothing on this site has ever been sold at any price, and so there is no “was” anywhere to strike through. Also directed for this instrument - from £6,000/yr org licences ### The commitment ladder One month less paid per year of service, per step, taken off this instrument's own anticipated annual rate of from £390/seat/yr — the first of the 2 figures directed for it, the rest in the two-SKU split table. The rate is held at the figure you sign for the whole term, so a multi-year commitment fixes the price as well as reducing it. Term | Months paid per year | What it means | Anticipated | Monthly billing | 12 | from £39/seat/mo × 12. Twenty per cent more than the annual term — that is the uplift for paying monthly, not a discount for paying annually. | from £468/seat/yr | 1-year term | 10 | Twelve months of service for the price of ten. This is the annual rate every band below is taken off. | from £390/seat/yr | 2-year term | 9 | Ten per cent off the annual rate, held at that figure for the whole term. | from £351/seat/yr | 3-year term | 8 | Twenty per cent off the annual rate, held at that figure for the whole term. | from £312/seat/yr | Three years for what two years of monthly billing costs Eight months paid per year, across three years, is 24 months paid for 36 months of service — one year in three carries no charge. On this instrument the three-year term totals from £936 per seat. That is 24 × £39 = £936 , and two years billed monthly is £468 × 2 = £936 . The same money. It buys three years instead of two, and the arithmetic is on the page so you can check it rather than take it. ### Seat bands This instrument is licensed per seat, so the seat count is a band. Bands are taken off the annual rate. Three of the twenty-one instruments are per-seat; on the rest a seat band would be a category error, and the scaling axis is entities instead. Seats | Band | Anticipated, per seat | 1–4 seats | The annual rate. | from £390/seat/yr | 5–9 seats | Ten per cent off the annual rate. | from £351/seat/yr | 10–19 seats | Twenty per cent off the annual rate. | from £312/seat/yr | 20+ seats | Past nineteen seats the licence is negotiated as a group licence rather than counted seat by seat. | Group licence, negotiated | The two bands multiply — they do not add A seat band and a term band are applied one after the other, not summed. Twenty per cent and twenty per cent is not forty per cent off; it is sixty-four per cent of the annual rate. Worked end to end on this instrument, at 12 seats on a three-year term: - **Annual rate per seat** — from £390 - **Seat band, 10–19 · ×0.8** — from £312 - **Term band, 3-year · ×0.8** — from £249.60 per seat per year - **× 12 seats** — from £2,995.20 per year - **Over the three-year term** — from £8,985.60 None of those figures is rounded to look tidy. A buyer who can reproduce the number has a reason to trust the vendor who published it, and a rounded arbitrary figure gives them nothing to reproduce. ### Commercial routes Pricing scales on seats and on term — bands from five seats, a group licence past nineteen, set out above with its own figures. [Licensing — seat bands, group licences and multi-year terms in full](https://hadalinstruments.com/licensing/) ### Support - Tier **Standard** at this instrument’s base contract — ticket, first response targeted at three business days . Support tier follows the annual contract value, not the price of a single unit — more seats, a suite licence or a group agreement raise the contract value and can raise the tier with it. [Support — the tiers, the ticket-only channel model, and what a target does and does not promise](https://hadalinstruments.com/support/) ### Trial mechanics The trial runs on a seat — thirty days, disclosed in plain words before you start it and cancellable in one step; none of it is live yet. [Trial mechanics in full, by delivery class](https://hadalinstruments.com/licensing/#trial). ### Read before you commit The documentation is published ahead of the product on purpose — intended behaviour is only a commitment if it exists first. Start with [installation and first run](https://hadalinstruments.com/docs/install/), then [the limits](https://hadalinstruments.com/docs/limits/): the conditions under which this instrument refuses to produce a number are the part worth reading before you pay. The full centre is at [/docs/](https://hadalinstruments.com/docs/), and the [support model](https://hadalinstruments.com/support/) states what a ticket does and does not cover. ## Questions and answers Answered from what this instrument publishes about itself. Nothing below is attributed to a customer, because there are none yet. ### Does this make an AI agent honest? No. Nothing does. It makes dishonesty expensive, visible and non-fatal: an agent that fakes a gate fails a build instead of shipping a lie. ### Is it a prompt file? No. It is a set of hard rails in CI that assume the agent will optimise for “task complete” over “task true”. Enforcement lives where the agent cannot bypass it. ### What counts as done under the receipt rule? A verifiable receipt: a passing build, an observed behaviour, or a content-hashed artifact. “The agent said it works” is not a state the pipeline can reach. ### Can I licence this instrument today? No. Nothing on this site is on sale — there is no checkout, no card capture, and no product account to create. Every figure on this page is an anticipated indication I have set so it can be read and compared, not an offer, and final pricing awaits my ratification. The launch list is the only thing you can join today. ## Change log NOT YET PUBLISHED Agent-Governance Constitution has not shipped, so there is nothing to record. When it does, every version lands here — dated, append-only, written by a person, and including the changes that removed a capability rather than added one. ## Where this sits Agent-Governance Constitution is one of the instruments in the [Honesty Stack suite](https://hadalinstruments.com/suites/stack/). How that suite measures — the per-instrument battery, and the receipts each measurement will carry — is set out in the [Honesty Stack methodology](https://hadalinstruments.com/methodology/stack/), part of the site-wide [measurement methodology](https://hadalinstruments.com/methodology/). ### Also in the Honesty Stack suite Agent-Governance Constitution shares the Honesty Stack suite with three other instruments. - P4 [Epistemic Harness](https://hadalinstruments.com/instruments/epistemic-harness/) Registry-as-a-service: pre-registration, walk-forward enforcement, the kill ledger. - P15 [Provenance-Grade Ingestion](https://hadalinstruments.com/instruments/provenance-grade-ingestion/) Vendors sell data. Nobody sells provenance discipline around data. - P19 [Fail-Visible Ops](https://hadalinstruments.com/instruments/fail-visible-ops/) In trading infrastructure, silence is a fault — not a status. ### Research behind this instrument Agent-Governance Constitution draws on two research notes on this site. - [What stops an AI agent from faking its results?](https://hadalinstruments.com/research/what-stops-an-ai-agent-from-faking-its-results/) Asked as: "what stops an ai coding agent from faking its results" - [The Burned Door: an AI agent destroyed its evidence](https://hadalinstruments.com/research/the-burned-door/) --- ## Raw artifact — NOT PUBLISHED FOR THIS PAGE No downloadable artifact ships with this page. Eight published measurements do, each content-hashed so a reader can verify the figures independently. Where a measurement is published here without one, that is a gap rather than a policy, and it is stated rather than left to be noticed. [Measurements that ship their data](https://hadalinstruments.com/research/) [Join the launch list](https://hadalinstruments.com/pricing/#waitlist)[How this suite measures](https://hadalinstruments.com/methodology/stack/)