Fix Cursor Chat Error calling tool edit_file Fast

Fix “Error calling tool 'edit_file'” in Cursor Chat

When working on large codebases in Cursor Chat, you might see “Error calling tool 'edit_file'” and the edit stalls. This is often caused by the model attempting oversized edits and not being explicitly instructed to call the correct tools.

Below are proven steps to fix it and keep edits moving.

Error Overview

What you’ll see

• Chat shows: Error calling tool 'edit_file'.
• Cursor appears stuck on “Generating,” especially on large edits.
• The assistant attempts to add very large diffs (hundreds of lines) in one go and fails.

Solution 1: Enable and clarify tool usage in your Project Rules

Why this works

Cursor needs explicit instructions and tool names in your Project Rules so the assistant knows it’s allowed to edit files and which tools to call. Without this, it may hesitate or fail to call edit_file correctly.

Steps

1. Open your Project Rules file in Cursor.
2. Add clear instructions that:
   • Identify the assistant’s role inside Cursor.
   • Permit viewing, editing, creating, and running files in the project directory.
   • Name the tools available and when to call them.
   • Instruct the assistant not to ask you to perform file actions manually.
3. Save the rules and retry your edit.

Sample Project Rules snippet

You are the AI used within the AI editor Cursor. You can view, edit, create, and run files within the project directory.

When asked to identify a bug, fix a bug, edit a file, or create a file, you must execute the appropriate tool:

- edit_file: Edit an existing file.
- create_file: Create a new file.
- read_file: Read the contents of a file.
- list_directory: List files and folders.
- gp_search: Search for code or terms across the project.

Do not ask the human to provide or create files; use the tools above to do the work yourself.

If a tool call fails (e.g., edit_file), ask for guidance and retry with smaller, incremental edits.

Solution 2: Instruct the model to edit in small chunks

Why this works

Large, single-shot edits are prone to timeouts and rate limits, especially on big files. Asking the assistant to edit in small chunks helps it apply changes gradually without failing.

Steps

1. Add a small-chunks rule to your Project Rules (near the top for priority):
   • “When editing files, make changes in small chunks to avoid large diffs.”
2. In your first chat message for the task, say:
   • “Please edit the file in small chunks.”
3. If you still see failures, ask the assistant to:
   • Limit each edit to a small section of the file.
   • Apply changes incrementally and wait for confirmation between edits.

Suggested wording to paste

Please edit the file in small chunks. Apply changes incrementally instead of large diffs.

If errors persist: start a fresh repo/session

Why this helps

Chats can carry state from earlier failures. Starting a fresh repository and session resets that state and avoids lingering issues. Clearing cache may not resolve it.

Steps

1. In GitHub Desktop (or your VCS), create a new repository and publish it (optional if local-only).
2. Open the new repo in Cursor.
3. Copy your project files into this fresh repo.
4. Re-add the updated Project Rules from Solution 1, including the small-chunks instruction from Solution 2.
5. Start a new chat and run the edits again.

Additional Tips

Keep edits manageable

• Ask the assistant to modify one section or function at a time.
• If a file is very large, split the task into multiple edits or separate files.

Be explicit about tool names

• If the assistant hesitates, say: “Use edit_file to apply the change.”
• Confirm the file path and section to edit to reduce back-and-forth.

Watch for rate limits

• If you see repeated failures, pause briefly and retry with even smaller changes.

Re-run with a clean first message

• Start the chat with:
  • The goal (what you want changed).
  • The file path(s).
  • “Please edit the file in small chunks using edit_file.”

Conclusion

This error usually comes from missing tool guidance and oversized edits. Fix it by updating your Project Rules to name and permit edit_file and related tools, and by instructing the assistant to make small, incremental changes. If the issue lingers, start a fresh repo and reapply these rules before continuing.