Skills System

Modular Skills System 🛠️

Gideon features a powerful “Skills” system that allows for rapid extension of its core capabilities. Every major feature in Gideon—from Voice AI to Threat Detection—is implemented as a skill.

What is a Skill?

A skill is a self-contained module that provides:

Commands: Specific CLI triggers (e.g., gideon brief).
Tools: Wrappers for LLM use via LangChain.
Lifecycle Hooks: Initialization and availability checks.

Built-in Skills

Skill	Description	Requirements
Security Research	Workflows for bug bounty and pentest assistance.	None
Threat Detection	Morpheus-powered analysis of network traffic and logs.	NVIDIA GPU
Data Analytics	RAPIDS-powered batch processing of IOCs.	NVIDIA GPU
Voice AI	Hands-free operation using PersonaPlex.	NVIDIA GPU
Governance	Audit logging and access control logic.	None
OpenClaw Sentinel	Sidecar security for OpenClaw AI agents. 8 commands covering gateway monitoring, skill scanning, prompt injection defense, hardening audits, credential protection, and memory integrity.	OpenClaw instance

Skill Structure

Skills are located in src/skills/. A typical skill directory looks like this:

src/skills/my-skill/
├── index.ts      # Main registration and metadata
├── tools.ts      # Tool definitions (LangChain)
├── commands.ts   # CLI command handlers
└── types.ts      # Skill-specific type definitions

Adding a New Skill

Adding a skill is a three-step process:

1. Define Metadata

Define what the skill provides and its hardware requirements.

2. Register Commands & Tools

Map your logic to the Gideon CLI and the Agent Core loop.

3. Register in the Registry

Export your skill from src/skills/index.ts.

[!TIP] Always implement a CPU fallback for skills that primarily use GPU acceleration (like NVIDIA NIM integrations) to ensure Gideon remains portable.