AI-Native SDLC

Anthropic’s playbook for rebuilding the software development lifecycle around agentic coding, from the Claude Academy course of the same name. One page for the model itself, then one page per play — what changes, how to start, governance, and how to measure it.

Start here

• The AI-Native SDLC — why the bottleneck moved out of the build phase, the six stages as a loop rather than a line, the committed artifact that carries work between them, and the split between advisory and deterministic controls.

Stage 1: Plan

• Capture as intent.md — the originator brainstorms the idea with Claude and commits it as a version-controlled proto-spec in their own terms, instead of handing it through backlog refinement.

Stage 2: Design

• Requirements and Design in One Session — Claude turns an approved intent.md into a spec.md constrained by the organization’s skills, flagging the policy conflicts an analyst would have escalated.

Stage 3: Build

• Plan Mode as the Default Starting Point — the plan becomes reviewable before any code exists, and the accepted plan.md is what later stages check the diff against. Also covers auto mode and the legacy source-of-truth problem.
• The CLAUDE.md as Institutional Memory — the onboarding document a new joiner would need, kept under a page and corrected whenever Claude makes the same mistake twice.
• Skills as Institutional Knowledge — encode one inconsistently-enforced policy as a skill, and understand it as an advisory control that needs a deterministic hook behind it.
• Parallel Sessions and Subagents — one engineer drives several sessions in separate worktrees, with recurring jobs packaged as subagents; the ceiling is review capacity.

Stage 4: Test

• Give Claude a Feedback Loop — the session verifies its own work before an engineer sees it, and the check itself is protected from the agent fixing the code.
• Continuous Evals in CI — 20 to 50 real tasks with checks, run whenever CLAUDE.md, skills, or hooks change, because configuration steers the agent and deserves regression testing.

Stage 5: Deploy

• AI in the PR Review Loop — every PR gets an identical set of review passes defined in REVIEW.md, and human attention moves up to intent and risk.
• Hooks as Approval Gates — each human approval the process requires expressed as a hook that can allow, ask, or block, with the regulated-enterprise managed settings file annotated key by key.
• CI/CD Integration and Deployment — Claude runs non-interactively in the pipeline for the judgment steps, sandboxed, with deploy and rollback exposed through MCP and autonomy tiered by environment.

Stage 6: Maintain

• Closing the Loop on Metrics — a deterministic detection script invokes Claude on a control-band breach, tiered by sigma, and the diagnosis is written back as an intent.md that restarts the lifecycle.