Inject good coding habits into AI agent framework directories.

Many AI coding agents load project config directories into their system context. Pandemonium finds those directories and appends curated habit prompts — so every agent session can start with clean-code principles already loaded.

Source: github.com/promptagainstthemachine/pandemonium

pandemonium.py --init
pandemonium.py --init --path .github
pandemonium.py --unicode --greek

How it works
Supported config layouts
Install
Usage
Examples
License

How it works

Reads English habit prompts from prompts/en/
Optionally translates (Google / DeepL / LibreTranslate)
Encodes text (format + charset encoders, including obfuscated --init hybrid)
Detects agent config directories under --path
Appends encoded prompts (idempotent — already-injected files are skipped)

Use --clean to strip injections and remove scaffold files Pandemonium created.

Supported config layouts

Pandemonium detects agents by directory markers, not product names.

Config root	Detection	Writable targets (examples)
`.claude/`	`.claude/` or `CLAUDE.md`	rules, memory, skills, agents
`.codex/`	`.codex/` or `AGENTS.md`	rules
`.cursor/`	`.cursor/` or `.cursorrules`	rules
`.github/`	copilot instructions	skills, agents
`.vscode/`	`.vscode/` / mcp.json	instructions, agents, skills
`.opencode/`	`.opencode/`	agents, commands, skills
`.gemini/`	`.gemini/`	skills, agents
Others	`.kiro/`, `.aider/`, `.windsurfrules`, `.openhands/`, `.continue/`, `.harness/`, `.pi/`, `.openclaw/`	per layout

Pass --include-home to also scan agent config under $HOME.

Install

With UV (recommended)

git clone https://github.com/promptagainstthemachine/pandemonium.git
cd pandemonium
uv sync
uv run pandemonium.py --help

With pip

pip install -e .
pandemonium --help

Requires Python ≥ 3.11.

Usage

`--init` (recommended)

Default workflow:

Reads prompts/en/habits.md
Encodes chunks with obfuscated hybrid encoders (no readable English in output)
Injects into detected config directories

pandemonium.py --init
pandemonium.py --init --path .github
pandemonium.py --init --clean --path .github

Useful flags

Flag	Description
`--path DIRECTORY`	Root to scan (default: current directory)
`--include-home`	Also scan `$HOME` agent configs
`--clean`	Remove injected blocks and scaffolds
`--unicode` / `--ascii` / `--invisible` / `--camelcase`	Format encoders
`--charset NAME`	Charset encoding (repeatable)
`--greek` / `--french` / …	Translate habit prompt language
`--backend google\|deepl\|libre`	Translation backend
`--list-charsets`	List charset encodings

Examples

# Default obfuscated hybrid injection
pandemonium.py --init

# Target only .github
pandemonium.py --init --path .github

# Explicit format + language
pandemonium.py --unicode --greek --path .

# Remove injections
pandemonium.py --clean --path .github

License

MIT — see the repository LICENSE file.

Table of contents