Skills

Extend what your agents can do with Agent Skills — portable, file-based packages of procedures, scripts, and reference material that agents discover and activate on demand. Think of them as plugins for your agent.

HiveMind OS implements the open Agent Skills standard. Where MCP provides callable tools, skills provide knowledge and procedures: how to approach a task, domain conventions, and scripted workflows.

Configuration

Skills are configured in the skills section of your config:

# ~/.hivemind/config.yaml
skills:
  enabled: true
  sources:
    - type: github
      url: https://github.com/org/skills-repo
  storage_path: ~/.hivemind/skills-cache

SkillsConfig Fields

Field	Type	Description
enabled	bool	Enable or disable skill discovery
sources	Vec<SkillSourceConfig>	List of skill sources (GitHub repos)
storage_path	Option<String>	Local cache directory for downloaded skills

Skill discovery scans configured sources and loads each SKILL.md frontmatter into a lightweight index. When a task matches a skill description, the full SKILL.md body is injected into the agent's context.

How Activation Works

1. Startup — HiveMind OS scans configured skill sources, loading each SKILL.md frontmatter (~100 tokens) into a lightweight index.
2. Match — When a task matches a skill description (keyword, semantic, or explicit request), the full SKILL.md body is injected into the agent's context.
3. Resources — Files in scripts/, references/, and assets/ load on demand as the agent follows instructions.

Creating Your Own Skill

1. Define the skill manifest

Every skill is a directory with a SKILL.md file. The YAML frontmatter is the manifest:

---
name: create-presentation
description: Build slide decks from data with branded templates
license: MIT
compatibility: ">=1.0"
metadata:
  author: Alice
  category: productivity
allowed_tools: "filesystem.*, shell.*"
---

## Instructions

1. Load the template from `references/TEMPLATES.md`
2. Generate charts using `scripts/generate_charts.py`
3. Assemble slides following brand guidelines in `assets/`

SkillManifest Fields

Field	Type	Description
name	String	Unique skill name
description	String	What the skill does
license	Option<String>	License identifier
compatibility	Option<String>	Version compatibility constraint
metadata	Map<String, String>	Arbitrary key-value metadata
allowed_tools	Option<String>	Tool glob patterns the skill may use

2. Add supporting files

Add scripts and reference files alongside SKILL.md:

my-skill/
├── SKILL.md
├── scripts/
│   └── generate_charts.py
├── references/
│   └── TEMPLATES.md
└── assets/
    └── slide-template.pptx

3. Publish (optional)

Share your skill by pushing the directory to a GitHub repo and adding it as a source in your SkillsConfig.

Skills + Data Classification

Skills integrate with HiveMind OS's security model. The data classification system applies to skills just as it does to conversations and connectors.

Classify Sensitive Skills

A skill that accesses external APIs or production systems should be used in sessions with appropriate classification (CONFIDENTIAL or higher). This prevents accidental data leakage through the skill's tool calls.

Agent Kits

Bundle personas, workflows, skills, and attachments into a portable .agentkit ZIP archive to share complete agent configurations with your team. See the Agent Kits guide for full details on exporting, importing, and namespace remapping.

Learn More

• Tools & MCP — How built-in tools and MCP complement skills
• Privacy & Security — Data classification and channel enforcement
• Personas Guide — Scoping skills and tools per persona
• Workflows Guide — Automating multi-step agent tasks