Apprentice for Openclaw

Apprentice enables programming by demonstration, allowing your AI agent to watch your actions and convert them into permanent, executable skills.

taha2053
v1.0.0
Feb 20, 2026
0
1.4k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install apprentice

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install apprentice 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 Apprentice?

Apprentice represents a significant leap in human-computer interaction by bringing Programming by Demonstration to AI agents. Instead of writing code or complex prompts, you simply perform a task while the agent observes your actions and listens to your intent. It bridges the gap between manual execution and automated workflows by identifying variables, constants, and logical steps in real-time.

Built as a local-first solution, Apprentice ensures that your workflows and observations stay on your machine. It synthesizes your demonstrations into durable Openclaw Skills that your agent can refine, chain, and execute forever. This approach captures the nuance of your specific work style, making your agent truly personalized to your professional environment.

Apprentice Use Cases

  • Automating repetitive development environment setups and project initializations.
  • Capturing complex multi-step deployment sequences that are unique to your infrastructure.
  • Training agents to handle administrative reporting tasks by showing them how you aggregate data.
  • Standardizing client onboarding processes through demonstrated file structures and git commands.
  • Creating a personalized library of daily routines that the agent can replay on command.

How Apprentice Works

  1. Activate observation mode by using trigger phrases like watch me or apprentice mode.
  2. Perform your task naturally, using CLI commands and file edits while narrating your logic to provide context.
  3. Signal the end of the demonstration using phrases such as done or stop watching.
  4. Review the synthesized workflow where the agent identifies the purpose, steps, variables, and constants.
  5. Approve the synthesis to save the new routine as a permanent skill in your workflows library.

Apprentice Setup

To get started with Apprentice, ensure the skill files are placed within your agent's directory. No external API keys are required as the synthesis happens within your existing LLM session.

# Navigate to your agent skills directory
cd your-agent/skills

# Clone the Apprentice repository
git clone https://github.com/Taha2053/apprentice

Once installed, the agent will recognize the trigger phrases automatically.

Apprentice Data Schema & Taxonomy

Apprentice maintains a structured hierarchy for every workflow it learns, ensuring that each demonstration is documented and reproducible. Each learned skill is stored in the workflows/ directory.

File Name Description
SKILL.md A fully valid Openclaw Skills definition that makes the workflow portable and editable.
run.sh The generated execution script containing the bash commands identified during observation.
observation.json A raw log of the observation session, allowing for manual auditing or re-synthesis.

Workflows are categorized by name and can be managed, edited, or deleted directly through the agent interface.

Apprentice Advanced Features

  • Workflow Chaining: Connect multiple learned skills to create complex, conditional automation pipelines.
  • OpenClaw Portability: Every learned workflow is a native skill that can be shared or published to other environments.
  • Intelligent Variable Detection: Automatically distinguishes between dynamic inputs like project names and static constants like template paths.
  • Narrative Understanding: Leverages LLM logic to understand why you are performing a step, not just what the step is.
  • Fully Local Synthesis: Processes all demonstrations on your machine to maintain strict privacy and security.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*