Claude AI Guide: Tips, Tutorials & Best Practices

Claude Code works with DeepSeek, GLM (Z.ai), and Kimi (Moonshot AI) models once you set ANTHROPIC_BASE_URL, each provider's API key, and the model name environment variables. The benefit is that you can swap only the model that answers while keeping the screen, tools, and settings, which helps when you want to cut costs or hit Claude's usage limits, though Anthropic does not support this use. The article draws on official documentation from each provider and Anthropic (checked October 19, 2026) to cover pros and cons, settings.json examples, per-million-token prices for nine models, estimated per-session costs, four factors that change billing, DeepSeek's search charges starting November 1, 2026 ($0.75 per 1,000 successful calls), comparisons with fixed-price plans such as GLM Coding Plan, and code privacy concerns such as DeepSeek's data storage in China.

Claude Haiku 5.5 (claude-haiku-5-5) is the lightweight model Anthropic released on October 7, 2026. For prompts up to 100,000 tokens it costs $0.10 input and $0.50 output per million tokens, one-tenth of Haiku 4.5. Above that it costs $0.50 input and $2.50 output, and tokens read from the cache count toward the length. It has a 1M-token context, 128K maximum output and a June 2026 knowledge cutoff, and it is the first Haiku that lets you choose the effort (medium by default). Based on official documentation and announcements checked against the originals on October 15, 2026, this article covers estimated costs for four models adjusted for the roughly 30% more tokens counted for the same text, how to read the benchmarks Anthropic published, how to choose between Sonnet 5.5 and Opus 5.5, how to set thinking and effort, request patterns that return a 400 error after moving from Haiku 4.5 and the changes that happen without an error, where haiku points in Claude Code (Haiku 5.5 on the Anthropic API from v2.1.293), and availability in the apps, GitHub Copilot and the cloud.

Claude's Max, Ultra, and Team plans come with credits you can use on the Claude API every month ($100 on Max, $200 on Ultra, and up to $500 in total on Team depending on the number of seats). However, you don't receive them until you link a Claude Console organization on the claude.ai billing page, and unused credits don't roll over. This article is based on my record of the screens when I claimed the credits on a Max 20x (now Ultra) plan on October 10, 2026, and on the official documentation, help articles, and terms, which I checked in the original on the same day. It explains where the credits can be used (the API, Message Batches, Managed Agents, the Agent SDK, the Playground, and claude -p run with an API key) and where they can't (interactive Claude Code, extra usage in the apps, Bedrock, and more); that linking partway through a cycle still gave me the full amount, expiring about three weeks later; that you can't change the linked organization yourself and an organization can receive only one plan's credits; the order credits are used in and auto-reload; how keeping ANTHROPIC_API_KEY set permanently switches Claude Code to API billing; ways to use the credits up and spend limits; and key points from the terms, such as the credits not being transferable or sellable.

Covers setting up investigation-only use in the Codex app, IDE, and CLI, and in Claude Desktop, VS Code, web, and mobile. Explains read-only versus Plan, where settings apply and persist, tool restrictions, ready-to-use prompts, and diff checks.

Claude Code mods (Claude Mods) are plugins that run JavaScript or TypeScript functions inside Claude Code, officially added in v2.1.287 on October 1, 2026. This article explains how a mod made of three files works, how it differs from traditional hooks written in settings.json, skills and MCP, what mods can do, such as drawing panes in the interface, creating commands that run without waiting and holding a tool call to ask the user, and limits such as run time, all based on the original text of the official docs checked on October 5 of the same year. A mod runs with your permissions outside the sandbox and can override permission decisions. A table shows that deny rules hold on Team or Enterprise or on a machine with managed settings, but on a personal plan a mod can approve even calls a deny rule refused. It also covers the results of reading the code of Anthropic's three official sample mods and counting which APIs they use, how to read claude plugin validate before installing, how to try mods, have Claude build one and turn them off, where mods run and the built-in mods, and a checklist before installing.

You can hand development to Claude Code cloud sessions without opening a new way into production: keep the code in a private GitHub repository, install the Claude GitHub App on that repository only, and have the production server just pull it with a read-only deploy key. This article is my record of starting to use cloud sessions on a separate development project on October 4, 2026, checked against the original text of the official docs as of that date. It explains that claude --cloud without GitHub sends the history of every branch plus uncommitted changes to tracked files (on Windows, regardless of file names), how to fix it when a different GitHub account is connected, "Only select repositories" and why public repositories appear in the list, why you should create a fresh environment per project since anyone who uses an environment can read its environment variables, that local files are invisible, and that Accept edits in the cloud stops at every command while Auto keeps going, all wrapped up in a pre-start checklist. It also shares a real case where a single instruction run autonomously used $226 of limited-time credit before I told it to stop midway, why credit drains so fast, and ways to cut costs.

Whether your Claude Code conversations are used to train AI models depends on your plan and your account setting. On the Free, Pro, and Max consumer plans, your data is used for training only while the model-improvement setting in claude.ai's privacy settings is on (kept for 5 years), and turning it off stops training (kept for 30 days). The setting belongs to your account, so there's no need to change settings files on the computer running Claude Code. Based on the original text of the Claude Code docs page "Data usage," Anthropic's Privacy Center, and the Claude Platform page "API and data retention" as of October 4, 2026, this article covers where the setting is and how to turn it off, the fact that opting out doesn't remove your data from past training, how 👍👎 ratings, /feedback, and conversations flagged as policy violations are handled, and training and retention by plan, including Team, Enterprise, the API, and Bedrock. It also explains that API inputs and outputs aren't used for training by default and are deleted within 30 days, what ZDR covers and why Covered Models such as Fable require 30 days of retention, the environment variables that stop usage metrics and error reports sent separately from training, the conversation transcripts stored in plaintext on your own computer, and how to choose a setup for confidential work.

Claude Tag is an Anthropic feature that lets you hand Claude research, document writing, recurring reports, and even draft GitHub pull requests just by typing @Claude in a Slack channel (announced June 23, 2026, in public beta). Only Team and Enterprise organizations can use it; on personal Pro and Max plans, the prompt may appear on claude.ai, but clicking it leads nowhere. I confirmed this is by design using the actual screen from clicking it on a personal Max plan and the matching row in the official troubleshooting page. This article lays out, from the original text of the official docs as of October 4, 2026, a decision flow for whether you can use it, how coworkers without a Claude seat can still ask in channels, where each kind of work is billed (the monthly limit options, and the 7-day, $50, 50-session limits on DMs from members without a connected account), how to ask in channels, DMs, and group DMs, channel memory, scheduled runs and channel monitoring with Routines, switching models, connecting tools with Access bundles and dedicated Claude accounts, the PR flow on GitHub and GitLab and two-person approval, how guests and Slack Connect are handled, data retention and deletion, common snags and commands, and how it differs from the earlier Claude in Slack, Cowork, and Claude Code.

Claude Code's /doctor prompt-audit (/checkup prompt-audit is the same) has Claude read CLAUDE.md, CLAUDE.local.md, AGENTS.md, and the rules, skills, commands, and subagents under .claude/, look for prompting written for older models, references to files or commands that don't exist, and instructions that contradict each other, and return a report with proposed diffs. Nothing changes until you ask, and it requires v2.1.283 or later. Based on the official docs and CHANGELOG as of October 3, 2026 and the audit guide bundled with Claude Code, this article covers how to run it and narrow it with a path, what it counts as stale (over-strong wording, missing paths, conflicts between files, and more), its rule against deleting context and prohibitions that exist for a reason, the six fields of the report and how to read confidence levels, and how it differs from /doctor, /claude-api prompt-audit, and the terminal claude doctor command. It also shares what happened when we ran it on this site's instruction files (7 of 9 findings applied, including a real drift where we forgot to rewrite an explanation after adding a rule) and caveats such as usage allowance, run-to-run variation, and the effort of checking findings before applying them.

Prompt caching reuses the computed results for the start of a request that matches an earlier one, so that part of the input costs less and runs faster. OpenAI (GPT-5.6 and later) turns it on by default and keeps the cache at least 30 minutes after last use (12 hours is optional), with writes at 1.25x the input price (2.5x for 12 hours) and reads at 0.1x (0.05x for GPT-6.1 Sol); Claude caches only when a request includes cache_control, with a 5-minute cache (1 hour is optional), writes at 1.25x (2x for 1 hour), and reads at 0.1x (0.05x for Opus 5.5 and Sonnet 5.5, 0.025x for Fable 5.1). Based on both companies' official documentation as of October 3, 2026, this article compares minimum length, breakpoints, sharing scope, and rate limits, works out how many reads it takes to break even, and explains common misses and how to check hits with usage and diagnostics.

Claude Code's /code-review is a bundled skill that reads your commits ahead of upstream plus uncommitted changes (or a PR, branch, or path you pass), hunts for correctness bugs, and reports them. /review is an alias, the review runs as a background subagent, and its cost counts toward your normal usage. This article covers how to pass a target, how to choose between low (only high-confidence findings) and medium to max (wider coverage, but you need to sort the results), the rule that leaving out the level reuses the last level you typed, the caveat that --fix edits made in the background can't be undone with /rewind, and where --comment posts on GitHub and GitLab. It also explains ultra, where multiple agents in the cloud return only verified findings (about 15 to 25 minutes; 3 one-time free runs on Pro and Max, then typically $5 to $25 per run in usage credits), and how /code-review differs from /simplify, /security-review, the GitHub App's Code Review, and claude-code-action, based on the original text of the official documentation and CHANGELOG as of October 2, 2026 (the level descriptions as of October 14). Finally, it shares the results of running /code-review high once on the code for this site's AI image checker tool (9 of 10 findings were real bugs, all fixed) and the downsides, including usage, the work of checking findings, and how to undo --fix.

On Windows, Claude Desktop refuses to start and shows "There's a problem with Claude. Reinstall the application from its original install location or contact your administrator." This is the message Windows shows when it refuses to launch a packaged (MSIX) app, and neither Anthropic's help center nor its documentation explains it or how to fix it (as of September 30, 2026). The dialog says to reinstall, but checking the state with Get-AppxPackage first tells you which route to take. If nothing is returned, the package registration is gone and a reinstall is all it takes. If Status reads Modified, NeedsRemediation, end the leftover processes and try Repair, then re-register with -DisableDevelopmentMode, then run the official installer over the top, and only as a last resort back up, remove and reinstall. A reinstall keeps your chat history and your Claude Code conversation logs (.claude\projects), but users report that the desktop app's session list was lost; claude --resume still opens those sessions. Microsoft's documentation limits Remove-AppxPackage -PreserveApplicationData to apps under development, and it has been reported to fail with 0x80073CFA. The article also sorts out the triggers seen in GitHub reports: crashes in the built-in browser, auto-updates while the app is running, and failed Cowork service registration.