ReferenceCurated resources
The sources this guide is built on — all from 2025–2026, the practitioners defining agentic coding as it's actually done. Grouped so you can go deeper on any section.
Matt Pocock — the workflow backbone
- Skills for Real Engineers — the canonical skill source (grill, PRD, issues, TDD, diagnose, review).
- The 7 Phases of AI-Driven Development — the full lifecycle.
- 5 Agent Skills I Use Every Day — the daily pipeline.
- My Skill Makes Claude Code GREAT at TDD — red-green-refactor with agents.
- Tracer Bullets: Keeping AI Slop Under Control — vertical slices.
- Ship Working Code While You Sleep (Ralph Wiggum) — the AFK loop.
- Your Codebase Is NOT Ready for AI — deep modules, small interfaces.
Anthropic — the engineering foundations
- Building Effective Agents — workflows vs agents, patterns over frameworks.
- Effective Context Engineering for AI Agents — the context window as a finite resource.
- Effective Harnesses for Long-Running Agents — agents across many context windows.
- How We Built Our Multi-Agent Research System — orchestrator/subagent architecture.
- Best Practices for Claude Code — explore/plan/code, verification loops, CLAUDE.md.
- Agentic Coding and Persistent Returns to Expertise — ~400k real sessions analysed.
- Towards Understanding Sycophancy in Language Models — why models drift toward agreement.
- How People Ask Claude for Personal Guidance — deference rises measurably under user pushback.
- Demystifying Evals for AI Agents — capability vs. regression evals, grading outcomes not paths.
- Finding Bugs with Claude and Property-Based Testing — real bugs in NumPy/SciPy/Pandas that example-based tests missed.
- Building a C Compiler with a Team of Parallel Claudes — why the automated task verifier has to be nearly perfect.
- Headless Mode and GitHub Actions integration — running Claude Code unattended inside CI.
- Code Review for Claude Code — multi-agent PR review; the 16%→54% substantive-review-rate figure.
Practitioners — planning, review, context, delivery
- Simon Willison — Writing Code Is Cheap Now — eight quality attributes for agent output.
- Addy Osmani — Agentic Code Review — risk-tiered review by blast radius.
- Addy Osmani — The Code Agent Orchestra — subagent orchestration; verification is the bottleneck.
- Birgitta Böckeler (Thoughtworks) — Maintainability Sensors for Coding Agents — automated non-functional checks.
- Birgitta Böckeler — Context Engineering for Coding Agents.
- GitHub — Automate Repository Tasks with Agentic Workflows — CI safety, human approval before merge.
- Every — Compound Engineering — the plan-work-review-compound loop.
- Jesse Vincent — Superpowers — skills + worktrees + subagents + TDD.
- Geoffrey Huntley — How to Build a Coding Agent — a from-scratch harness primer.
- Agentic AI Foundation — AGENTS.md — the open, cross-tool standing-instructions standard.
- Thoughtworks — Technology Radar — the adopt/trial/assess/hold framework for reviewing tools and dependencies over time.
- Chrome for Developers — Carousels with CSS — a concrete example of a platform feature newer than most models' training data.
- Google — Modern Web Guidance — the Baseline-aware skill pack referenced above; and Chrome DevTools for agents (announced 2025-09).
- Harry Brignull — deceptive.design — the canonical dark-pattern catalogue.
- Google SRE — Canarying Releases — the reference pattern for staged rollout and automated rollback.
- MSR 2026 — Are Coding Agents Generating Over-Mocked Tests? An Empirical Study — 1.2M commits; agents mock more than humans do.