Awesome-Agentic-Engineering CLAUDE.md

CLAUDE.md

This file guides Claude Code when working in this repository.

This repository is a public, maintained awesome list for agentic AI engineering, not an application codebase. There is no application build or test suite to run for normal review tasks; advisory Markdown and link checks exist and are documented in CONTRIBUTING.md and AGENTS.md. The README.md is the product.

Claude Code should read this file first, then use AGENTS.md as the shared repository operating protocol.

North Star

• Preserve README.md as the canonical public artefact.
• Keep the list selective, durable, technically useful, neutral, and easy to scan.
• Apply the editorial stance to every decision: reliability over novelty · evaluation over intuition · architecture over tooling · systems thinking over prompt engineering.
• Help the maintainer make fast, consistent, low-friction decisions.
• Prefer small, precise edits over broad rewrites.
• Do not broaden the list beyond agentic engineering and the adjacent technical areas already represented in the README.

Claude's Role

Claude may assist with:

• PR review and rubric scoring
• Issue triage
• README and appendix entry review
• Broken-link investigation
• Duplicate detection
• Section placement
• Neutral description rewrites
• Maintainer comment drafts
• Small safe maintainer edits when explicitly asked
• Improvements to agent instruction files when asked

Claude must not:

• Add entries without checking scope, the hard gates, evidence, link quality, duplicates, and placement
• Invent facts about a resource
• Preserve promotional claims
• Add ranking, pricing, novelty, adoption, or performance claims without strong evidence
• Rewrite the taxonomy without explicit instruction
• Edit unrelated files
• Touch protected areas unless instructed
• Ask contributors to make trivial fixes the maintainer can safely make

Repository Facts

• AGENTS.md contains the full tool-agnostic operating protocol.
• CONTRIBUTING.md contains contributor-facing rules, the evidence policy, and the PR process.
• RUBRIC.md contains the seven-dimension scoring rubric; ANTI-PATTERNS.md lists what fails it; appendix/benchmark-and-evidence-policy.md sets source order and evidence tags.
• .github/ISSUE_TEMPLATE/ and .github/PULL_REQUEST_TEMPLATE.md contain public contributor guidance.
• README.md contains introductory content, a Table of Contents, infographics, and the main curated sections; appendix/ holds overflow and fast-moving lists.
• CHANGELOG.md records every content change as a bullet under ## [Unreleased].
• The main list uses both bullet entries and tables, plus a deep-dive template for major entries. Match the surrounding section exactly.
• Each entry goes in its single best-fit section; do not mirror an item across sections without a justified distinction.
• New categories should normally be handled separately.
• docs/ and assets/github/ are generated site output — do not hand-edit.