Wiki

AI-assisted development: technical reference

Last updated 26 October 2026Tool covered Claude Code (CLI)

Lookup pages for engineers who use, or are about to adopt, an AI coding assistant in a development workflow: commands, configuration, checks and open questions. Pages are short, dated and added over time.

01 CLI operations in four tiersCommands grouped by use: enter and orient, daily work, context and configuration, non-interactive runs in CI. 25 October 2026 02 Shortcuts and habitsInterrupt, mode-switch and clear-context keys; naming files instead of describing them; verifying with real commands. 25 October 2026 03 Tool behaviour to know in advanceBranches and commits it may create, a working directory that may not be the one you opened, and answers that miss the real question. 25 October 2026 04 Configuration worth having.gitignore entries, what CLAUDE.md should contain, and which permission mode fits which kind of work. 25 October 2026 05 Checklist before putting AI into a CI pipelineFeedback loop, independent oracle, frozen tests, iteration cap, pinned model, stored transcripts, no skip annotations. 25 October 2026 06 Questions to ask before deploying in an enterprise environmentTen groups of questions to answer, and write down, before an AI coding assistant is used on shared code. 25 October 2026 07 The gateway model for AI access in an enterpriseWhat a gateway controls and what it does not, what is actually sent, trade-offs, cost effects, and how to tell which environment you are in. 26 October 2026

Conventions

  • Commands are written as they are typed. A command that depends on your version, platform or account is marked, and the page points to the tool's own help instead of guessing.
  • The list of commands, flags and permission modes changes between versions. Each page states which version it was checked against; the source of truth is /help and claude --help in the session you are running.
  • Anything not verified is labelled as such.
  • No page names a specific organisation, system or data set. Principles are written at a general level.

Unofficial notes, not affiliated with or endorsed by Anthropic. Command names, flags and behaviour change between versions; the official documentation and the tool's own help output are authoritative.