Agents Onboarding

Agents-Onboarding

Overview

Create AGENTS.md at repository root to enable fresh agent context to apply all repository standards and external skills. Fresh agents (no conversation history) must understand how to work in the repository by reading only AGENTS.md.

REQUIRED SUB-SKILL: superpowers:brainstorming REQUIRED SUB-SKILL: superpowers:verification-before-completion

When to Use

• User integrates AI agents into repository for first time
• User asks to set up agent guidance, rules, or onboarding documentation
• Repository is missing agent-specific onboarding docs (no AGENTS.md)
• Fresh agent context needs to understand repository standards without conversation history
• Existing AGENTS.md is incomplete or outdated and needs enhancement

Core Workflow

1. Announce skill and why it applies.
2. Analyze repository for existing standards (brownfield) or define new (greenfield).
3. Create AGENTS.md at repository root with required sections.
4. Validate fresh context test: agent with only AGENTS.md can apply standards.
5. Commit with evidence.

AGENTS.md Required Sections

• Repository Overview: Purpose, architecture, key constraints
• Required Skills: External (superpowers:*) and custom skills with triggers
• Development Standards: Code style, testing, review, deployment
• Repository Conventions: Directory structure, naming, documentation
• Process Guidance: PR workflow, deployment, escalation paths
• Context Optimization: What agents must know, focus areas

Red Flags - STOP

• "Agent can figure it out"
• "We'll document standards later"
• "Just need basic agent help"
• "Only one person uses agents here"

All of these mean: Apply skill, create comprehensive AGENTS.md.

References

• references/AGENTS-TEMPLATE.md - Complete template with annotations
• references/standards-discovery.md - Repository analysis techniques
• references/rationalizations.md - Excuses and reality checks