cd /news/ai-tools/ailock-hide-source-from-ai-assistant… · home topics ai-tools article
[ARTICLE · art-19195] src=github.com pub= topic=ai-tools verified=true sentiment=· neutral

AiLock: Hide source from AI assistants while tests still run

AiLock released a new encryption tool that keeps source code encrypted on disk while allowing developers to run Python programs normally in memory. The tool encrypts files in place so AI coding assistants and file-reading tools see only ciphertext, while `ailock run` decrypts code inside the runtime process without writing plaintext back to the working tree. This memory-only decryption approach aims to prevent AI assistants from accessing proprietary code during development.

read5 min publishedMay 31, 2026

Keep code encrypted on disk, decrypt it only in memory, and still run it normally.

AiLock encrypts files in place so filesystem-level AI access (read_file

, grep

, cat

, codebase indexing) sees only binary ciphertext. At the same time, developers can run encrypted Python code, import encrypted modules, read encrypted data files, and edit locked files through controlled plaintext views. The central idea is memory-only decryption: plaintext is materialized inside the AiLock runtime process, not written back to the working tree.

Disk: ciphertext for AI and ordinary file readers. Runtime: plaintext only inside the controlled execution process.

Most encryption tools protect files at rest, but make the code unusable until it is decrypted back onto disk. AiLock is built for a different workflow:

AI-opacity: coding assistants that read the working tree see ciphertext.** Memory-only execution**:ailock run

decrypts encrypted Python files inside the process and executes them without restoring plaintext on disk.Transparent imports: encrypted modules can import each other.** Transparent file I/O**:open()

,Path.read_text()

, andPath.read_bytes()

can return plaintext inside the runtime.GUI plaintext viewport:ailock open

lets the developer inspect and edit files without leaving plaintext in the working tree.Recovery path: encrypted backups and optional recovery keys help recover damaged or forgotten-password files.

  • Python 3.11 or newer pip

  • Runtime packages installed automatically from pyproject.toml

:argon2-cffi

,cryptography

, andpyzipper

tkinter

forailock open

; it is bundled with many Python installations, but some Linux distributions package it separately aspython3-tk

Install from GitHub:

git clone https://github.com/lo2589/AILOCK.git
cd AILOCK
pip install .

For editable development installs:

git clone https://github.com/lo2589/AILOCK.git
cd AILOCK
pip install -e .

Check the command:

ailock --help

If the command is not on your PATH

, use the module entry point:

python -m aloc --help
ailock lock secret.py

cat secret.py
grep "password" .

ailock show secret.py
ailock run secret.py
ailock open .

ailock unlock secret.py

The key idea:

ailock lock app.py      # app.py becomes ciphertext on disk
ailock run app.py       # app.py is decrypted in memory and executed

ailock run

is the core feature. It decrypts the entry file in memory, executes the plaintext inside the Python process, and leaves the working-tree file as ciphertext. No plaintext copy is written next to the encrypted file.

ailock run main.py
ailock run -m mypackage
ailock run app.py -- --port 8080

While the program is running, AiLock installs hooks so application code can behave as if the files were plain:

encrypted .py on disk -> decrypt in memory -> exec/import inside Python
encrypted data file   -> decrypt in memory -> open()/Path.read_text()

Inside your program, no AiLock-specific code is required:

import json
from secret_module import algorithm

with open("config.json") as f:
    config = json.load(f)

print(algorithm(config))

If secret_module.py

or config.json

is locked, AiLock decrypts it for the runtime while the filesystem still contains ciphertext.

Encrypt a file or directory in place.

ailock lock secret.py
ailock lock src/
ailock lock secret.py --recovery

Notes:

  • Directories are processed recursively.
  • Already locked files are skipped.
  • Plaintext backups are stored as encrypted ZIP backups under .ailock/backups/

by default. --recovery

prints a recovery key. Save it separately; it is not shown again.

Run encrypted Python code without writing plaintext back to disk.

ailock run main.py
ailock run -m mypackage
ailock run app.py -- --port 8080

Runtime interception layers:

  • import hook for encrypted Python modules

  • patched builtins.open

  • patched pathlib.Path.read_text

andpathlib.Path.read_bytes

Open a GUI plaintext viewport/editor for a directory.

ailock open .
ailock open src/

Locked files are decrypted for display. Saving writes encrypted content back to disk.

Print decrypted content to stdout without modifying the file.

ailock show secret.py
ailock show secret.py | head

Decrypt a file or directory back to plaintext on disk.

ailock unlock secret.py
ailock unlock src/ --backup

Recover a locked file using a recovery key generated by --recovery

.

ailock recover secret.py

Start a stdin/stdout JSON-RPC workspace server for controlled plaintext access.

ailock freelock .

Example requests:

{"method": "list_files", "params": {}, "id": 1}
{"method": "read_file", "params": {"path": "main.py"}, "id": 2}
{"method": "grep", "params": {"pattern": "TODO"}, "id": 3}
{"method": "write_file", "params": {"path": "main.py", "content": "..."}, "id": 4}
{"method": "flush", "params": {}, "id": 5}
ailock status file.py
ailock forget
ailock forget --all
ailock config
ailock config backup-dir /path/to/backups
ailock init --as aa

ailock init --as <name>

installs a local launcher under a custom command name. This is useful when you want the unlock command to be deployment-specific.

AiLock targets filesystem-level AI access. It is designed for coding assistants and indexers that inspect files through ordinary reads. In that model, locked files reveal only ciphertext.

AiLock does not claim to stop a fully informed local adversary who can run arbitrary commands, capture process memory, or trick the user into decrypting files. For stronger isolation, combine AiLock with operating-system execution policy, process isolation, and careful secret handling.

  • Argon2id for password-derived keys
  • ChaCha20-Poly1305 for authenticated encryption
  • independent random file keys
  • password wrapping for file keys
  • optional recovery-key wrapping
  • encrypted ZIP backups for emergency recovery
aloc/
  cli.py        command-line interface
  runner.py     in-memory execution engine
  workspace.py  decrypted workspace API and JSON-RPC handler
  gui.py        tkinter GUI editor
  crypto.py     Argon2id and ChaCha20-Poly1305 helpers
  format.py     locked-file format parser/encoder
  fileops.py    atomic writes and backup helpers
  cache.py      sudo-style password cache
  manifest.py   .ailock manifest and backup management
  recovery.py   recovery key support
  install.py    custom command-name launcher

argon2-cffi

cryptography

pyzipper

tkinter

for the GUI, provided by many Python installations

MIT

── more in #ai-tools 4 stories · sorted by recency
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/ailock-hide-source-f…] indexed:0 read:5min 2026-05-31 ·