Claude Code Common Workflows: A Practical Guide for Developers

A practical guide to Claude Code's most useful everyday workflows, from exploring codebases and fixing bugs to running parallel sessions, resuming conversations, and piping Claude into scripts.

---

You just cloned a repo you've never seen before. No README worth reading, no senior dev to ask, and a deadline that isn't moving. Or maybe you're mid-debug, your context is a mess, and you're not even sure where the bug started.

Claude Code was built for exactly these situations. It works like a pair programmer that lives in your terminal, reads your files, runs your tests, and keeps up with your project the way a good teammate would.

This guide cuts straight to the workflows you'll actually use every day, with the exact prompts and commands that make them work.

---

Explore an Unfamiliar Codebase

When you're new to a project, start broad and drill down.

cd /path/to/project
claude

Then ask:

give me an overview of this codebase

explain the main architecture patterns used here

how is authentication handled?

Once you know the lay of the land, find specific code:

find the files that handle user authentication

trace the login process from front-end to database

Tips:

• Use the project's own domain language in your prompts
• Reference files directly with @ to skip the back-and-forth: Explain the logic in @src/utils/auth.js

---

Fix Bugs Faster

Share the error, get suggestions, apply the fix.

I'm seeing an error when I run npm test

suggest a few ways to fix the @ts-ignore in user.ts

update user.ts to add the null check you suggested

The key is giving Claude enough context. Tell it the command that reproduces the issue, the stack trace if you have one, and whether the error is consistent or intermittent. The more specific you are, the better the fix.

---

Refactor Legacy Code

Claude can find outdated patterns and modernize them safely.

find deprecated API usage in our codebase

suggest how to refactor utils.js to use modern JavaScript features

refactor utils.js to use ES2024 features while maintaining the same behavior

run tests for the refactored code

Always do refactoring in small increments and run tests between changes. Ask Claude to explain the benefits before you commit to an approach.

---

Write and Run Tests

Claude reads your existing test files and matches your project's style automatically.

find functions in NotificationsService.swift that are not covered by tests

add tests for the notification service

add test cases for edge conditions in the notification service

run the new tests and fix any failures

Claude is especially good at spotting the edge cases you might miss: error conditions, boundary values, and unexpected inputs. Ask it to check for those specifically.

---

Create Pull Requests

You can do this in one shot or step by step:

create a pr for my changes

Or guide it manually:

summarize the changes I've made to the authentication module

create a pr

enhance the PR description with more context about the security improvements

Once a PR is created with gh pr create, the session links to it automatically. To return later: claude --from-pr <number>.

---

Add or Update Documentation

Claude follows your existing doc style (JSDoc, docstrings, etc.) without you having to spell it out.

find functions without proper JSDoc comments in the auth module

add JSDoc comments to the undocumented functions in auth.js

improve the generated documentation with more context and examples

check if the documentation follows our project standards

---

Reference Files and Directories with @

Use @ to include files or directories instantly, without Claude having to read around looking for them.

Explain the logic in @src/utils/auth.js

What's the structure of @src/components?

Show me the data from @github:repos/owner/repo/issues

File paths can be relative or absolute. You can reference multiple files in one message: "Compare @file1.js and @file2.js". Directory references give you a file listing, not file contents.

---

Work with Images

Claude Code handles visual context too. Drop in a screenshot, diagram, or mockup and reference it in your prompt.

Ways to add an image:

1. Drag and drop into the Claude Code window
2. Copy an image and paste with Ctrl+V (not Cmd+V)
3. Provide the path: Analyze this image: /path/to/your/image.png

Then ask:

Here's a screenshot of the error. What's causing it?

Generate CSS to match this design mockup

What HTML structure would recreate this component?

---

Resume Previous Conversations

When a task spans multiple sessions, don't start over. Claude saves every conversation locally.

claude --continue

This picks up your most recent session in the current directory. To choose from a list:

claude --resume

Or use /resume from inside a running session.

---

Run Parallel Sessions with Worktrees

Work on two things at once without your changes colliding. Each worktree is a separate checkout on its own branch.

claude --worktree feature-auth

Run the same command with a different name in a second terminal to start an isolated parallel session:

claude --worktree bugfix-login

---

Use Plan Mode Before Editing

If you want to review changes before they touch your files, switch to plan mode. Claude reads everything and proposes a plan, but makes no edits until you approve.

claude --permission-mode plan

You can also press Shift+Tab mid-session to toggle into plan mode.

---

Delegate Research to Subagents

When exploring a large codebase, all those file reads fill your context fast. Hand the investigation off to a subagent instead.

use a subagent to investigate how our auth system handles token refresh

The subagent does its work in its own context window and hands you back just the findings.

---

Schedule Claude on a Recurring Basis

Claude Code can run on a schedule so it handles routine tasks without you lifting a finger. Here's how the options compare:

Option	Where it runs	Best for
Routines	Anthropic-managed infrastructure	Tasks that run even when your machine is off
Desktop scheduled tasks	Your machine via the desktop app	Tasks needing access to local files or uncommitted changes
GitHub Actions	Your CI pipeline	Tasks tied to repo events like opened PRs
/loop	Current CLI session	Quick polling while a session is active

When writing prompts for scheduled tasks, be explicit about what success looks like. The task runs on its own and can't ask for clarification. For example: "Review open PRs labeled needs-review, leave inline comments on issues, and post a summary in the #eng-reviews Slack channel."

---

Pipe Claude into Scripts

Claude works like any Unix tool. Use it non-interactively for CI, pre-commit hooks, or batch processing.

git log --oneline -20 | claude -p "summarize these recent commits"

Stdin and stdout work as expected. This is useful for anything you want to automate without opening an interactive session.

---

Q&A

1. Can I use Claude Code in a non-code project like a notes vault or documentation folder?

Yes. Claude Code works in any directory. Point it at a folder of markdown files and it can search, edit, and reorganize content the same way it handles code.

2. What's the difference between claude --continue and claude --resume?

--continue picks up the most recent session in your current directory automatically. --resume opens a list so you can choose which past session to return to.

3. How does Claude handle test generation for different languages or frameworks?

Claude reads your existing test files first and matches the style, framework, and assertion patterns already in use. You don't need to specify them.

4. What does plan mode actually prevent?

In plan mode, Claude reads your files and proposes what it would change, but makes no edits until you give the go-ahead. It's useful when you want to review the scope of changes before anything is written to disk.

5. Can I use Claude Code for scheduled tasks even when my computer is off?

Yes, if you use Routines. Those run on Anthropic-managed infrastructure and can trigger on a schedule, API calls, or GitHub events. Desktop scheduled tasks require your machine to be running.

6. What happens to the context when I use a subagent?

The subagent runs in its own context window. It reads files and does the investigation there, then hands back a summary to your main session. Your main context stays clean.

7. Is the @ file reference different from just asking Claude to read a file?

Using @ is faster. It immediately includes the file's full content in the conversation without Claude needing to use a tool call to read it. It also adds relevant CLAUDE.md files from parent directories.

8. How do worktrees prevent conflicts when running parallel sessions?

Each worktree is a separate checkout on its own branch. Changes in one worktree don't affect the other because they're working from different copies of the repo.

9. Can I pipe output from other tools into Claude, not just git?

Yes. Any tool that writes to stdout can be piped into Claude. Examples: linting output, test results, log files, or any command-line output you want Claude to analyze or summarize.

10. How does Claude know what PR format or documentation style my project uses? Claude reads your existing files. For PRs, it looks at previous PR descriptions if available. For docs, it reads existing comments and docstrings to match the format before generating new ones.

References

• Claude Code Common Workflows