Configs

AI Coding Configuration Guides

Turn project knowledge into instructions an agent can discover, follow, and verify without hiding important rules in chat history.

Good configuration is specific to the repository: real commands, directory boundaries, permission limits, validation steps, and a definition of done.

AGENTS.md

Configure global, root, nested, and override scope for commands, architecture, boundaries, and completion checks.

Understand discovery, precedence, size limits, nested scope, and validation.

Copy an annotated repository instruction file.

Choose project instruction files for Codex, Claude Code, Cursor, and mixed-agent repositories, with a practical content template.

Configure project, user, local, nested, imported, and path-scoped Claude Code guidance.

Choose locations, imports, rules, and verify loaded context.

Copy a concise Claude Code project file.

A repeatable setup using Cursor for navigation and review, Claude Code for scoped tasks, and git worktrees for isolation.

Choose Always, intelligent, file-scoped, manual MDC rules, or simpler AGENTS.md instructions.

Create valid .mdc files, globs, activation modes, and verification checks.

Verify the editor and repository before adding rules.

Choose stdio or HTTP and make access, secrets, tool approval, and verification explicit.

Set transport, scope, credentials, tool policy, and validation for coding agents.

Diagnose discovery, startup, auth, and zero-tool failures.

Decide what must stay local, where local models are good enough, and when an approved cloud agent is worth the tradeoff.

Continue by task

Build useful AGENTS.md, CLAUDE.md, task briefs, review checklists, and repository instruction templates without generic filler.

Practical setup paths for Claude Code, Codex, Cursor, Copilot, local models, and controlled AI coding workflows.

Diagnose AI coding setup failures, missing context, Cursor indexing problems, unsafe output, and production readiness issues.