Skip to main content
Documentation menu

CARL v1.0 keeps eight published pillars. v1.1 adds deltas to all eight and drafts six more. Pillars are not weighted against each other; a subject must make progress across the catalog to advance.

Deltas to published pillars

These pages extend v1.0 BLD/CQV/TST/TIR/SEC/DLM/DOC/OBS with 1xx criterion IDs. Several checks use scope: portfolio so a multi-application subject is scored as one product, not as a pile of unrelated apps.

PillarPrefixRegistry codeWhat it measures
Build & EnvironmentBLD-bldPortfolio toolchain, secret-free proof, slice deploys.
Code Quality & ValidationCQV-cqvShared quality config, import graph, forbidden patterns.
TestingTST-tstHonest coverage, package contracts, fake/staging split.
Task Intake & RoutingTIR-tirApplication-scoped intake, shared labels, owner routing.
Security & Access ControlSEC-secPath-attributed secrets, per-slice audit, deploy identity.
Delivery & MeasurementDLM-dlmAffected-slice CI, isolated previews, portfolio DORA.
Documentation & KnowledgeDOC-docNested agent context, portfolio map, classified env.
ObservabilityOBS-obsShared telemetry schema, trace propagation, per-slice SLO.

New pillars with draft criteria

PillarPrefixRegistry codeWhat it measures
Cost & FinOpsCOST-costToken, infrastructure, and operating-cost visibility.
Agent GovernanceAGV-agvAutonomy bounds, approval gates, and policy-as-code.
CompositionCOM-comModule boundaries and published composition contracts.
DataDAT-datData contracts, classification, retention, and access.
DesignDSN-dsnDesign tokens, accessibility, and UI primitives.
ResilienceRES-resFailure handling, backup, restore, and degradation.

Why com instead of cmp

The v1.1 ticket lists this pillar as CMP. The app already uses cmp for the v1.0 Compliance scorecard pillar. v1.1 Composition therefore uses registry code com and criterion prefix COM-. Compliance cmp is unchanged.

The six draft pillars