A subtraction-first engineering agent skill designed to write the minimum correct code needed by actively fighting over-engineering, bloat, and dependency creep.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install minimalist
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install minimalist using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Minimalist?
The Minimalist skill transforms an AI agent into a rigorous subtraction engineer, based on the philosophy that code is a liability that occasionally pays rent rather than an asset. It forces the system to evaluate every line written, prioritizing deletions, native features, and standard libraries over new code and unnecessary external dependencies. The skill operates under the belief that every line written is a line that must be permanently maintained, tested, and secured.
By establishing a strict process called "The Descent," this agent skill systematically evaluates whether requirements can be eliminated via YAGNI (You Aren't Gonna Need It), satisfied via current codebase reuse, or solved via built-in platform capabilities before any new logic is introduced. It provides a reliable guard against speculative engineering, ensuring your codebase remains clean, fast, and maintainable without ever sacrificing security or safety boundaries.
Minimalist Use Cases
- Refactoring and Technical Debt Reduction: Streamlining bloated systems by identifying and deleting unrequested abstractions, dead code, and redundant helpers.
- Preventing Dependency Creep: Evaluating tasks to use native platform features, standard libraries, or already-installed packages instead of pulling in new third-party dependencies.
- Symptom vs. Root Cause Bug Fixing: Analyzing code blast radiuses to implement a single shared guard instead of applying multiple redundant hotfixes across individual callers.
- Enforcing YAGNI Principles: Automatically pushing back on speculative requirements or future-proofing scaffolding that isn't required for the immediate task.
- Writing Highly Concise Code: Driving code generation during high-intensity periods where the diff is the deliverable and prose explanation needs to be kept to an absolute floor.
How Minimalist Works
- Problem Analysis: The skill triggers upon identifying keywords like "minimalist", "yagni", or "trim it", initiating a full analysis of the problem, targeted files, and caller graphs.
- The Descent Evaluation: The agent evaluates the task against a prioritized 7-step optimization ladder, stopping at the very first step that fully solves the problem.
- Abstraction Stripping: The agent actively strips out unrequested architectures, such as single-implementation interfaces, single-product factories, or unused configuration options.
- Guardrail Check: Before code finalization, the system runs an uncompromised validation check to ensure security, error handling, authentication, and input sanitization remain completely intact.
- Rejection Logging: If an explicit scope or external dependency is rejected, the skill outputs a one-line notification to the user and records the decision details into a local rejection script.
Minimalist Setup
To enable the Minimalist agent functionality within Openclaw Skills, include the skill configuration in your active context.
You can dynamically adjust or query the operational intensity during a session using the following command structures:
# Change intensity level to ultra for maximum code compression
/minimalist ultra
# Return to the standard configuration
/minimalist full
# Check measured savings or metrics gathered by the agent
/minimalist-gain
# Temporarily deactivate the subtraction constraints
/minimalist off
Minimalist Data Schema & Taxonomy
The skill manages its optimization workflow through active logging scripts and state levels configured within the runtime environment.
Metric and Logging Actions
When the agent rejects a dependency or architectural scope, it verifies the repository structure and records the event:
| Action / File | Description | Execution Context |
|---|---|---|
node scripts/log-rejection.js |
Logs rejected approaches, step positions, and estimated lines of code saved. | Executed from the repo root if the script is present. |
/minimalist-gain |
Displays measured optimization metrics and real savings recorded during the session. | User-invoked command payload. |
Intensity Level State Configuration
- lite: Advisory mode. Applies the descent ladder and flags over-engineering while deferring to user-specified structures.
- full: Default mode. Enforces all subtraction rules, pushes back on speculative scope in a single line, and implements the baseline operational floor.
- ultra: Hard-minimal output mode. Delivers highly compressed code changes accompanied by a maximum of three lines of prose explanation.
Minimalist Advanced Features
- The Descent Protocol: A strict 7-layer architectural filter ranging from complete requirement deletion (YAGNI) down to minimal new code implementation.
- Automated Rejection Tracking: Seamlessly integrates with local repository logging tools (
scripts/log-rejection.js) to log architectural decisions and track lines of code saved. - Absolute Safety Guardrails: Strict operational boundaries that prevent the optimization engine from ever stripping out input validation, auth checks, rate limits, error handling, resource cleanup, or test configurations.
- Multi-Skill Coexistence Engine: Intelligently yields prose-style choices to external output-terseness skills while maintaining its internal code-minimalism constraints without conflicting instructions.
- Language-Aware Explanations: Preserves structural syntax, articles, particles, and requested translations for non-English and mixed-language inputs to prevent loss of semantic meaning.
SKILL.md
Loading
Related Openclaw Skills
Minimalist Audit
Audit your codebase to automatically identify and rank deletion candidates like dead code, unused dependencies, and redundant helpers.
Minimalist Gain
An AI agent skill to measure and report codebase simplicity by tracking avoided lines of code, rejected scope, and declined dependencies.
Podcast Video Publisher
Converts podcast audio and artwork into high-quality 1080p MP4 slideshows and publishes them directly to structured YouTube playlists with automatic error and rate-limit recovery.
PixelVault List
A powerful CLI skill designed to list recently uploaded images and retrieve direct CDN URLs or rich JSON metadata from PixelVault.
QA QuickCheck
An AI-powered daily testing assistant that automates static code analysis, dynamic API testing, and regression analysis for your pull requests.
English Reading Analyzer
An advanced sentence-dissection and reading comprehension assistant that breaks down complex English sentences using a rigorous 5-step linguistic methodology.
Featured*
