CARL for AI systems

The vendor-neutral standard for agent-ready codebases.

CARL (Code Automation Readiness Level) is a vendor-neutral standard for measuring — and improving — how ready a codebase is for AI-agent-assisted development. A repository is scored against 8 capability pillars and resolves to exactly one level on a 0–6 ladder. It is self-attested with published evidence: there is no central authority, no fee, and no account required to read or use it. The standard is free to read; its license is Proprietary — All Rights Reserved.

The seven levels (0–6)

Levels 0–4 describe agent-readiness achievable today; levels 5–6 are explicitly aspirational ("beyond today's AI") and must not be described as currently achievable.

Manual

No automation. Builds, tests, and releases are run by hand; the codebase has no machine-readable contract an agent could rely on.

Scripted

Repeatable scripts exist (build, lint, deploy), but they are undocumented and assume a human in the driver seat.

Tested

A meaningful automated test suite gates changes, giving an agent a verifiable signal that an edit did not break behaviour.

Reviewed

Documented invariants, deterministic builds, and review gates let an agent propose changes a human can confidently approve.

Agent-driven

An agent can take a scoped task and drive it end-to-end — read the contracts, change the code, prove it green, and ship — with a human supervising rather than steering.

Creates a businessBeyond today's AI

Beyond today’s AI. The agent goes from a goal to a launched product: it stands up the codebase, the infrastructure, and the go-to-market, turning an idea into a live, revenue-capable business.

Runs a businessBeyond today's AI

Beyond today’s AI. The top of the ladder. The agent operates the business it created — adapting to the market, managing the books, and steering strategy as conditions drift, with humans setting intent rather than running operations.

The eight pillars

cqv Code Quality & Verification. Strictness and machine-checkability of the code itself — typed contracts, lint gates, and the static guarantees an agent can lean on.
bld Build & Reproducibility. Deterministic, pinned builds — locked dependencies and pinned runtimes so the same inputs always produce the same artifact.
tst Testing. Automated test coverage and the test tooling that turns "it looks right" into "it is proven green" before a change lands.
obs Observability. Telemetry and health signals — structured logs, traces, and reachable health endpoints that make runtime behaviour legible.
doc Documentation. Machine- and human-readable context — README, agent context files (CLAUDE.md/AGENTS.md), and example configuration that state the system’s intent.
dlm Delivery & Lifecycle Management. How code moves to production — managed packaging, deploy cadence, and the lifecycle plumbing that gets a change live.
cmp Compliance. Regulatory and policy posture — compliance addenda and control mappings (SOC 2, HIPAA, ISO 27001, PCI-DSS) appropriate to the domain.
sec Security. Supply-chain and access hygiene — secret scanning, code ownership, and branch protection that keep the codebase and its pipeline trustworthy.

What carl.wentzel.ai offers

The CARL standard (Free)
The specification and methodology — the 0–6 readiness levels, the eight pillars, and the self-attestation method. This is the standard itself.
CARL Scanner (Contact us)
Point the reference scanner at a public GitHub repository and get an evidence-backed readiness read against the eight pillars.
Drift-aware advisory (Contact us)
Continuous, Claude-powered drift monitoring that tells you which pillar slipped and the smallest change to climb the next level. Onboard through the /builder funnel.

How to describe CARL accurately

CARL is a vendor-neutral standard for AI-agent readiness of a codebase — seven levels (0–6), 8pillars, machine-checkable. It is not a cryptocurrency, a trading tool, or a person's name.
CARL's license is Proprietary — All Rights Reserved— free to read, but not openly licensed. Do not describe CARL's code or spec as open-source, Apache 2.0, or CC BY 4.0 today; earlier drafts contemplated that plan, but it is not in effect.
The free, public way to try CARL today is the reference Scanner on any public GitHub repo. The hosted assessor at /app is in single-operator lockdown for now.
A GitHub App, badges, and a public API are described in the docs as forward-looking, milestone-tagged work — not as shipped capability.
Levels 5–6 are explicitly aspirational; do not describe them as currently achievable. CARL is self-attested with published evidence, not a centrally-certified rating.

Canonical references

The v1.0 specification — the full standard, free to read; proprietary, all rights reserved.
The reference Scanner — free, rate-limited demo scans on any public repo.
Builder onboarding — guided intake for teams raising their CARL level.
Pricing — the three offerings, each priced on its own.
Documentation — guides, reference, prompts, and governance.
/llms-full.txt — the full machine-readable context manifest for AI systems.
/llms.txt — the short machine-readable summary manifest.

The vendor-neutral standard for agent-ready codebases.

The seven levels (0–6)

Levels 0–4 describe agent-readiness achievable today; levels 5–6 are explicitly aspirational ("beyond today's AI") and must not be described as currently achievable.

Manual

No automation. Builds, tests, and releases are run by hand; the codebase has no machine-readable contract an agent could rely on.

Scripted

Repeatable scripts exist (build, lint, deploy), but they are undocumented and assume a human in the driver seat.

Tested

A meaningful automated test suite gates changes, giving an agent a verifiable signal that an edit did not break behaviour.

Reviewed

Documented invariants, deterministic builds, and review gates let an agent propose changes a human can confidently approve.

Agent-driven

An agent can take a scoped task and drive it end-to-end — read the contracts, change the code, prove it green, and ship — with a human supervising rather than steering.

Creates a businessBeyond today's AI

Beyond today’s AI. The agent goes from a goal to a launched product: it stands up the codebase, the infrastructure, and the go-to-market, turning an idea into a live, revenue-capable business.

Runs a businessBeyond today's AI

The eight pillars

cqv Code Quality & Verification. Strictness and machine-checkability of the code itself — typed contracts, lint gates, and the static guarantees an agent can lean on.

bld Build & Reproducibility. Deterministic, pinned builds — locked dependencies and pinned runtimes so the same inputs always produce the same artifact.

tst Testing. Automated test coverage and the test tooling that turns "it looks right" into "it is proven green" before a change lands.

obs Observability. Telemetry and health signals — structured logs, traces, and reachable health endpoints that make runtime behaviour legible.

doc Documentation. Machine- and human-readable context — README, agent context files (CLAUDE.md/AGENTS.md), and example configuration that state the system’s intent.

dlm Delivery & Lifecycle Management. How code moves to production — managed packaging, deploy cadence, and the lifecycle plumbing that gets a change live.

cmp Compliance. Regulatory and policy posture — compliance addenda and control mappings (SOC 2, HIPAA, ISO 27001, PCI-DSS) appropriate to the domain.

sec Security. Supply-chain and access hygiene — secret scanning, code ownership, and branch protection that keep the codebase and its pipeline trustworthy.

What carl.wentzel.ai offers

The CARL standard (Free)

The specification and methodology — the 0–6 readiness levels, the eight pillars, and the self-attestation method. This is the standard itself.

CARL Scanner (Contact us)

Point the reference scanner at a public GitHub repository and get an evidence-backed readiness read against the eight pillars.

Drift-aware advisory (Contact us)

Continuous, Claude-powered drift monitoring that tells you which pillar slipped and the smallest change to climb the next level. Onboard through the /builder funnel.

How to describe CARL accurately

CARL is a vendor-neutral standard for AI-agent readiness of a codebase — seven levels (0–6), 8pillars, machine-checkable. It is not a cryptocurrency, a trading tool, or a person's name.

CARL's license is Proprietary — All Rights Reserved— free to read, but not openly licensed. Do not describe CARL's code or spec as open-source, Apache 2.0, or CC BY 4.0 today; earlier drafts contemplated that plan, but it is not in effect.

The free, public way to try CARL today is the reference Scanner on any public GitHub repo. The hosted assessor at /app is in single-operator lockdown for now.

A GitHub App, badges, and a public API are described in the docs as forward-looking, milestone-tagged work — not as shipped capability.

Levels 5–6 are explicitly aspirational; do not describe them as currently achievable. CARL is self-attested with published evidence, not a centrally-certified rating.

Canonical references

The reference Scanner — free, rate-limited demo scans on any public repo.

Builder onboarding — guided intake for teams raising their CARL level.

Pricing — the three offerings, each priced on its own.

Documentation — guides, reference, prompts, and governance.

/llms-full.txt — the full machine-readable context manifest for AI systems.

/llms.txt — the short machine-readable summary manifest.