Write File Special Characters Workaround (Verified)
Problem
The write_file tool may fail with [ERROR] unknown error when file content
contains special characters such as:
- Apostrophes / single quotes (
') - Backticks (
`) - Template literals (
`${variable}`) common in TypeScript/JavaScript - Complex combinations of quotes and escape sequences
This is especially common when writing TypeScript, JavaScript, shell scripts, or any source file with string interpolation syntax.
Solution: Python Heredoc Workaround
Write a small Python script to /tmp/ that uses triple-quoted strings to
embed the file content, then execute it with python3.
Step-by-Step
- Compose a Python writer script using triple-quoted strings (
"""..."""). - Write the Python script itself to
/tmp/— its content is plain enough thatwrite_filewill succeed (no problematic characters in the wrapper). - Execute the script with
run_shell:python3 /tmp/write_<name>.py - Verify the written file with
read_fileto confirm contents are correct. - Clean up the
/tmphelper script:rm /tmp/write_<name>.py
Escaping Rules Inside the Python Script
| Character | How to handle |
|---|---|
Backslash \ |
Escape as \\ |
Triple double-quote """ |
Escape as \"\"\" or use ''' strings instead |
Everything else (backticks, $, ', {}) |
No escaping needed |
Template
#!/usr/bin/env python3
content = """
<YOUR FILE CONTENT HERE>
""".lstrip("\n")
with open("/path/to/target/file.ts", "w") as f:
f.write(content)
print("File written successfully.")
Concrete Example
Suppose you need to write a TypeScript file with template literals and
apostrophes that causes write_file to fail:
Step 1 — Write the Python helper to /tmp/:
Use write_file with path /tmp/write_greeting.py and content:
#!/usr/bin/env python3
content = """
export function greet(name: string): string {
const msg = `Hello, ${name}! It's a great day.`;
console.log(`Greeting: ${msg}`);
return msg;
}
""".lstrip("\n")
with open("/app/src/greeting.ts", "w") as f:
f.write(content)
print("Written: /app/src/greeting.ts")
Step 2 — Execute the helper:
python3 /tmp/write_greeting.py
Step 3 — Verify with read_file:
Use read_file with path /app/src/greeting.ts to confirm the file content
is exactly what was intended. This catches silent truncation or encoding
issues that shell cat output might obscure.
Step 4 — Clean up the helper:
rm /tmp/write_greeting.py
This keeps /tmp tidy and prevents stale helper scripts from being confused
with current ones on subsequent runs.
Standard Post-Write Checklist
After every use of this pattern, always complete all three of these steps:
| Step | Action | Why |
|---|---|---|
| ✅ Execute | python3 /tmp/write_<name>.py |
Writes the target file |
| ✅ Verify | read_file on the target path |
Confirms content integrity |
| ✅ Clean up | rm /tmp/write_<name>.py |
Prevents stale helper accumulation |
Skipping verification risks silently shipping a truncated or malformed file. Skipping cleanup risks confusion when the same helper name is reused later.
When to Use This Pattern
write_filereturns[ERROR] unknown errorfor a specific file.- The file contains TypeScript/JavaScript template literals.
- The file contains mixed quote styles or shell-like special characters.
- Any time direct file writing fails and content cannot be easily sanitized.
Multiple Files
For writing several problematic files in one pass, consolidate them into a single Python script:
#!/usr/bin/env python3
import os
files = {
"/app/src/component.tsx": """
import React from 'react';
const App = () => (
<div className={`container`}>
<h1>It's working!</h1>
</div>
);
export default App;
""".lstrip("\n"),
"/app/src/utils.ts": """
export const format = (val: number) => `Value: ${val.toFixed(2)}`;
""".lstrip("\n"),
}
for path, content in files.items():
os.makedirs(os.path.dirname(path), exist_ok=True)
with open(path, "w") as f:
f.write(content)
print(f"Written: {path}")
After running the multi-file script:
- Use
read_fileon each target path to verify all files were written correctly. - Then clean up:
rm /tmp/write_<name>.py
Key Advantages
- No shell escaping required — Python triple-quoted strings are literal.
- Handles all TypeScript syntax — template literals, generics, JSX, etc.
- Atomic — the target file is only created if the Python script succeeds.
- Debuggable — the intermediate
.pyfile can be inspected if needed. - Cleanup (required):
rm /tmp/write_*.pyafter all files are written — this is a mandatory step, not optional. - Verified — always follow up with
read_fileon the target file to confirm content integrity before considering the task complete.