Kieran Klaassen

Overview

Kieran Klaassen is a creator highlighted for demonstrating a structured workflow around Claude Code, centered on his Compound Engineering plugin. In the newsletter references, he shows how AI coding sessions can be turned into repeatable operating systems using slash commands such as `workflows plan`, `workflows work`, `assess`, and `triage`. The core idea is a planning–coding–assessing–codifying loop that captures learnings locally and feeds them back into future runs.

For AI Product Managers, this matters because Klaassen’s workflow points to a practical pattern for making AI-assisted development more reliable over time. Rather than treating each coding session as isolated, his approach creates a persistent memory layer through local documentation, architecture decisions, and reusable instructions. That makes Claude Code outputs more consistent, testable, and aligned with team conventions—important qualities for PMs managing AI-enabled product delivery.

Key Developments

• 2026-02-09 — Kieran Klaassen is featured demonstrating how to make Claude Code “better every time you use it” through the Compound Engineering plugin. The workflow uses slash commands like `workflows plan`, `workflows work`, `assess`, and `triage` to create an iterative loop for planning, implementation, evaluation, and codification. The mention also notes that the plugin stores learnings as Markdown in `/docs/architecture-decisions/` and `/docs/solutions/`, updates `claude.md` so rules are injected into future planning prompts, and pairs Claude Code with Opus 4.5 and Playwright to generate end-to-end browser tests and PR artifacts. It further references the alias `CC="claude code --dangerously-skip-permissions"` to suppress permission prompts for unattended sessions.
• 2026-02-10 — Klaassen is again mentioned for demonstrating the Compound Engineering plugin for Claude Code CLI. The focus remains on the same planning–coding–assessing–codifying loop, emphasizing how iterative insight capture in a local docs directory can continuously improve generated code quality over repeated usage.

Relevance to AI PMs

• Designing repeatable AI workflows: Klaassen’s approach offers a concrete pattern for turning ad hoc AI coding into a system with explicit stages—planning, execution, assessment, and codification. AI PMs can use this model to standardize how teams interact with coding agents.
• Creating product memory for AI tools: The use of local docs, architecture decisions, and root instruction files shows how teams can preserve lessons learned and inject them into future agent runs. PMs can apply this idea to improve consistency across features, repos, and releases.
• Operationalizing autonomous QA and delivery: The use of Playwright, generated browser tests, logs, recordings, and unattended CLI sessions illustrates how agent workflows can support testing and pull request creation. For PMs, this is relevant when evaluating how far AI can be trusted in delivery pipelines and where safeguards are needed.

Related

• compound-engineering-plugin — Klaassen’s featured plugin, used to structure Claude Code workflows and capture codified learnings.
• claude-code — The CLI environment where his workflow patterns are demonstrated.
• peter-yang — Appears alongside Klaassen in the referenced content, indicating a contextual connection in the newsletter feature.
• playwright — Used in the demonstrated workflow for auto-generated end-to-end browser tests, UI interaction, logging, and video artifacts.
• compound-engineering — The broader system or framework associated with the plugin and workflow methodology Klaassen demonstrates.