Guides

Claude Code production guides.

Evergreen references for turning Claude Code into an engineering practice.

Use these guides when you need the production shape: MCP servers, skills, hooks, context, reliability, governance, security, and team adoption.

Operating model

Start with the operating model

Begin here if you need the operating model before choosing stack, governance, or behavior details.

Cornerstone Guide Start here

Running Claude Code as a Production Engineering Practice

How a consultancy delivers billable client engineering with Claude Code, gated by deterministic validators and demonstrated through published receipts.

Last reviewed May 22, 2026 Read the guide

Full index

All production guides

The full collection stays visible in static HTML so readers, crawlers, and AI systems can inspect the complete set.

Cornerstone Guide Agent Observability in Production: The Trace Is the Evidence Why a production agent's run has to be captured as external telemetry, what a trace actually records, how you read one to locate a failure, and where observability stops and evaluation, provenance, and governance begin. Last reviewed June 16, 2026 Read guide Cornerstone Guide Claude Model Change Management in Production: Treating a Model Version as a Dependency You Don't Control A Claude model version is a production dependency you don't set the release cadence of. The repeatable discipline for every model event, capability upgrade, deprecation, forced migration, silent behavior shift, or sudden unavailability: baseline, canary on your own traces, route by model and effort, and keep a continuity path. Last reviewed June 15, 2026 Read guide Cornerstone Guide Agent Autonomy Gates in Production: Calibrating What an Agent May Do Without You Agent autonomy is a calibration, not a trust switch. What permission modes buy, why both all-human and all-structural gates fail on the evidence, and how teams widen the autonomous lane without losing the brakes. Last reviewed June 12, 2026 Read guide Cornerstone Guide Cost-Aware Agentic Design in Production: The Bill Is a Design Readout, Not a Meter Token spend in an agentic system is set by five design choices: model tier, effort, cache discipline, context shape, and fan-out width. This guide treats cost as an architecture property, covering the June 2026 pricing mechanics, the lever order that preserves quality, the attribution stack that separates telemetry estimates from the metered invoice, and the ROI arithmetic that only works once spend is attributed to outcomes. Last reviewed June 11, 2026 Read guide Cornerstone Guide Refusal Handling and Model Fallback in Production: The Failure Your Dashboards Read as Success Claude Fable 5 returns safety-classifier refusals as successful HTTP 200 responses, so error-rate monitoring never sees them. This guide treats refusal handling and model fallback as a production discipline: the stop_details anatomy, the three retry paths and where each one is available, the observability contract built on usage.iterations, and the policy decisions no SDK default can make for you. Last reviewed June 10, 2026 Read guide Cornerstone Guide Claude Code Team Adoption in Production: The Rollout Is the Work Buying Claude Code licenses is not adoption. Individual speedup does not become organizational delivery, and trust, not tooling, gates real usage. This guide diagnoses why team adoption stalls after the license buy, then runs the rollout as an engineering practice you build: redesign the work, measure behavior over activation, standardize the practice so it survives the champion, and re-earn the skeptic with one good session. Last reviewed June 9, 2026 Read guide Cornerstone Guide The Agentic Test Pyramid in Production: Where Each Check Belongs When the Apex Is Also an Agent The classic test pyramid stratifies tests by scope and its top layer is still deterministic. The agentic test pyramid stratifies checks by determinism, and its apex is itself non-deterministic. That inversion changes where every check belongs, because you cannot make the apex reliable, only push more of the work down to a base that can be argued with by nobody. Last reviewed June 7, 2026 Read guide Cornerstone Guide Agentic Security in Production: Threat-Modeling an Agent That Cannot Tell Instructions From Data A tool-using agent is a new trust boundary, because the model cannot separate instructions from data and becomes a confused deputy the moment untrusted content reaches it. Agentic security is not input validation; it is containment by design: name the lethal-trifecta legs, scope what the agent can reach, gate the side-effecting actions, and assume any single layer can be breached. Last reviewed June 2, 2026 Read guide Cornerstone Guide Context Engineering for AI Coding Agents in Production: Why the Window Is the Unit of Work A coding agent in a large codebase is governed less by the model and the prompt than by what is in its context window when it acts. Context engineering is the discipline of curating, retrieving, compacting, and persisting that window, and a bigger window is not a substitute for it. Last reviewed June 1, 2026 Read guide Cornerstone Guide Agentic AI Governance in Production: Who Owns the Bar When the Agent Ships A written policy sets the rules but cannot enforce them on an autonomous agent. Agentic AI governance is the deny-by-default gates, scoped permissions, provenance records, and named verification-bar owners that put those rules in the execution path, between an agent and a shipped artifact. Last reviewed May 31, 2026 Read guide Cornerstone Guide Claude API in Production: A Runtime, Not a String Function, and What It Leaves to You The Claude API is a structured runtime, not a text-in/text-out endpoint. It hands you capability primitives but leaves conversation state, cost, and verification of non-deterministic output to you. Owning that boundary is the production work. Last reviewed May 31, 2026 Read guide Cornerstone Guide Claude Code Hooks in Production: The Gate the Model Doesn't Get to Skip Why a hook is the layer of an agentic coding stack the model cannot choose to skip, why a fail-open hook can be worse than no hook, and where enforcement belongs. Last reviewed May 31, 2026 Read guide Cornerstone Guide Agent Reliability in Production: A Verification Loop, Not a One-Time Test Why you cannot unit-test a non-deterministic agent, why reliability is the inverse of your known failure modes, and how the verification loop has to keep re-running, in parallel, as the system evolves. Last reviewed May 29, 2026 Read guide Cornerstone Guide Claude Code Skills in Production: Two-Axed Discoverability and the Patterns That Make Skills Compound A Claude Code skill earns its value through two-axed discoverability, the same module answering a slash command AND a natural-language request. Designing for both surfaces deliberately is the work; catalog hygiene is the pattern that lets skills compound past the threshold where description-routing accuracy starts to decay. Last reviewed May 25, 2026 Read guide Cornerstone Guide MCP Servers in Production: Ownership Inversion, the Protocol Vocabulary Limit, and What the Spec Leaves to You An MCP server moves the integration to a system into the server and out of the agent. The inversion pays inside a finite protocol vocabulary; outside it, the protocol is overhead. The spec also draws a second boundary worth naming, where functional integration moves but governance does not. Last reviewed May 25, 2026 Read guide Cornerstone Guide Subagent Orchestration in Production: Trade-offs and Failure Modes Why isolation per worker is the load-bearing property of subagent orchestration, where the pattern stops paying, and how the orchestration layer itself fails. Last reviewed May 23, 2026 Read guide Cornerstone Guide Running Claude Code as a Production Engineering Practice How a consultancy delivers billable client engineering with Claude Code, gated by deterministic validators and demonstrated through published receipts. Last reviewed May 22, 2026 Read guide

Related surfaces

Related authority surfaces

Use these surfaces when you need vocabulary, field notes, or a buyer path instead of another guide.

Glossary Definitions for the terms used across the guide collection. Blog Field notes and production lessons that expand the guide collection. Services The buyer path when you want help applying the guides.

Want to talk about how this applies to your team?

Book a Free Intro Call

Not ready for a call? Take the free AI Readiness Assessment instead.