Standards

An agent is only as good as what it can reach.

Most descriptions of AI development talk about tools. This page describes the systems the work actually lives in, the standards we hold that work to, and the one system an agent never touches.

  • Twelve systems, each with a defined access pattern
  • Two standards published in full
  • Secrets stay human-only, by design

A project is twelve systems, not one codebase

Most diagrams of software delivery draw phases. Phases are not the problem. The problem is that the information an agent needs is scattered across a dozen systems, and an agent is only as useful as the ones it can actually reach.

Each system below has a defined access pattern. Retrieval means the agent queries an index rather than reading raw files. Protocol means a structured connection to a service. Filesystem and command line mean direct access. One system is closed entirely.

SystemWhat lives in itHow an agent reaches it
DocumentationProject handbook, product vision, technical solution, decisions, and the domain knowledge that explains how the client's business worksRetrieval
SpecificationSpecs, acceptance criteria, backlog, prioritization, and delivery stateCommand line or protocol
DesignWireframes, interfaces, interaction flows, tokens, and component documentationProtocol or command line
DevelopmentSource code, configuration, branch rules, and the committed context file every agent reads at the start of a sessionFilesystem and command line
ValidationTest scenarios per acceptance criterion, coverage, and the traceability between themFilesystem and command line
DeliveryThe pipeline that builds and deploys, and the distribution that ships releasesCommand line
InfrastructureEnvironments, infrastructure as code, and the rules that provision and constrain agentsCommand line
ObservabilityError tracking, performance, structured logging, health checks, alert routing, and agent telemetryCommand line or protocol
CommunicationChannels, notifications, system events, and schedulingProtocol or command line
GovernanceAI usage policy, client disclosure, data classification, tool vetting, and the agent identity and permission modelRetrieval and filesystem
AccountContracts, budget, delivery metrics, and escalations. Role-gated, not open to the team by defaultRetrieval, restricted
SecretsAPI keys, credentials, environment variables, and signing certificatesNo access

Domain knowledge is the one most often missing

If a client's business is specialized, and in health services, payments, or industrial work it always is, an agent without that context produces confident wrong answers. Any document explaining how the business actually works belongs in the documentation system, whoever wrote it.

The system an agent never reaches

Secrets is human-only. Not restricted, not logged, not role-gated. Closed.

An agent references a variable name and never sees a value. That rule is the reason the system exists as its own domain rather than as a folder inside infrastructure: its job is to define how credentials are named, where they live, and how they are injected, so that none of them can arrive in a prompt.

Access patterns are easy to write down and easy to quietly widen. This is the one that never widens.

What the standards cover

Six standards cover the twelve systems.

StandardCovers
SpecificationWhat gets built, and how completely it is described before it is built
ValidationWhat proves the build matches the spec, designed before the build starts
DeliveryThe quality gates a change passes before it reaches production
DevelopmentThe build loop, and what belongs in the repository before an agent works in it
ObservabilityThe layers that turn a silent runtime failure into a signal
DesignHow design decisions survive the trip into code

Where you stand

Knowing the standards is not the same as knowing your own position against them. The AI readiness audit scores four dimensions on evidence, tags every finding by how strongly it is supported, and returns a ranked gap list. It is the same procedure we run on our own projects.

The failure this is built to prevent

An agent given an unclear requirement does not slow down. It does not ask a question or check with a designer. It implements a plausible reading of what it was given, completely and confidently.

That produces a specific failure. One pass generates the implementation and the tests from the same reading of the spec. When the reading is wrong, the code is wrong and the tests agree with it. Nothing surfaces.

Every standard below exists to close that gap. The specification removes the ambiguity before the build. The validation plan is written from the spec rather than from the finished code, so a test asks whether the work does what was required instead of whether it does what it does.

Want this run against your project?

The readiness audit scores four dimensions against these standards and returns a ranked gap list with the evidence behind each finding.