Can claude -p edit files?

Yes, if you allow it. In print mode nobody can answer permission prompts, so actions that need approval are denied. Add --allowedTools "Read,Edit" or --permission-mode acceptEdits to let Claude Code change files.

How do I get only the text answer from JSON output?

Pipe the JSON into jq and read the result field: claude -p "Summarize this project" --output-format json | jq -r '.result'. The same JSON also holds the session id, which you can pass to --resume later.

How can a script tell if claude -p failed?

Check the exit code. Claude Code exits with 0 on success and with a non-zero code when the run fails. An invalid flag is reported on stderr before the run starts.

How to Run Claude Code from the Command Line

Install the Claude Code CLI
Run a prompt with claude -p
Useful claude -p options for scripts
Related guides

You can run Claude Code from the command line without opening an interactive session. Add -p (short for --print) to the claude command to use print mode, also called headless or non-interactive mode: Claude Code takes one prompt, prints the answer and exits. Use it in scripts, cron jobs or for a quick question.

Install the Claude Code CLI

On macOS, Linux or WSL, run the installer:

curl -fsSL https://claude.ai/install.sh | bash

On Windows, run this in PowerShell:

irm https://claude.ai/install.ps1 | iex

You can also install it with Homebrew:

brew install --cask claude-code

Or with npm:

npm install -g @anthropic-ai/claude-code

Then start Claude Code once and sign in in the browser. Claude Code needs a Pro, Max, Team or Enterprise plan, or a Claude Console account.

claude

Run a prompt with claude -p

Put the prompt in quotes after claude -p:

claude -p "what is the purpose of life? answer one word"

Here is what Claude Code print mode looks like:

Claude Code running claude -p with a one-word prompt in the terminal and printing the answer Love

Claude Code prints only the answer, with no header, and returns you to the shell. Markdown is not rendered, so **Love.** is bold text in Markdown.

Interactive session or print mode:

Diagram: need a back-and-forth chat? Run claude. Otherwise run claude -p with a prompt

Useful claude -p options for scripts

Send a file or command output to Claude. Print mode reads piped text, and you can save the answer with >:

cat build-error.txt | claude -p "explain the root cause of this build error" > output.txt

How input and output flow:

Diagram: a prompt and piped text go into claude -p, and the answer goes to the terminal, a file or JSON

Get JSON for a script. The answer is in the result field, next to the session id and cost:

claude -p "summarize this project" --output-format json

Let Claude change files without asking. List the tools it may use:

claude -p "run the tests and fix any failures" --allowedTools "Bash,Read,Edit"

Allow file edits only:

claude -p "apply the lint fixes" --permission-mode acceptEdits

Continue the last conversation with a follow-up prompt:

claude -p "now explain it in one sentence" --continue

For all flags, run claude --help or see the Claude Code headless mode docs.

How to Run Codex from the Command Line, the same task with codex exec.
Claude Code YOLO Mode: What The Flag Means, before you turn off permission prompts.