AI SummaryDefines standardized conventions for creating, structuring, and managing AI agents across `.claude/agents/` and `.opencode/agent/` directories, ensuring consistency, security, and maintainability in multi-agent projects. Developers working with AI agents in monorepos, enterprise systems, and DevSecOps workflows benefit from clear role definitions and tool permission standards.
Install
Copy this and paste it into Claude Code, Cursor, or any AI assistant:
I want to set up the "AI Agents Convention" agent in my project. Please run this command in my terminal: # Add AGENTS.md to your project root curl --retry 3 --retry-delay 2 --retry-all-errors -o AGENTS.md "https://raw.githubusercontent.com/wahidyankf/open-sharia-enterprise/main/governance/development/agents/ai-agents.md" Then explain what the agent does and how to invoke it.
Description
Standards for creating and managing AI agents in .claude/agents/ (primary) and .opencode/agent/ (secondary) directories
AI Agents Convention
This document defines the standards for creating, structuring, and managing AI agents in the .claude/agents/ directory (primary source of truth) and .opencode/agent/ directory (auto-generated). Edit .claude/agents/ first, then sync to .opencode/. All agents must follow these conventions to ensure consistency, maintainability, and proper integration with the project.
What are AI Agents?
AI agents in this project are specialized AI assistants defined in the .claude/agents/ directory (source of truth), with auto-generated OpenCode configuration in .opencode/agent/. Each agent has: • Specific expertise in a particular domain or task • Defined tool permissions limiting what operations it can perform • Clear responsibilities to avoid overlap with other agents • Integration with project conventions through references to AGENTS.md and convention documents • Unlimited token budget - Focus on execution quality, not token usage (reliable compaction mechanism handles context management)
Why We Need Agent Conventions
Without standards, agents can become: • Inconsistent in structure and quality • Overlapping in responsibilities, causing confusion • Insecure through tool permission creep • Unmaintainable as the project grows This convention ensures all agents are: • PASS: Well-structured and documented • PASS: Single-purpose and focused • PASS: Secure through explicit tool permissions • PASS: Consistent with project standards
Scope
This convention applies to: • All agent files in .claude/agents/ (primary) and .opencode/agent/ (secondary) • References to agents in AGENTS.md • Agent validation rules in repo-governance-checker
Discussion
Health Signals
My Fox Den
Community Rating
Sign in to rate this booster