Skip to content
Agent

Agent Definition File Specification v3.0

by klappe-pm

AI Summary

A structured YAML/JSON specification framework for defining AI agents with consistent governance, role definitions, and capabilities across Claude platforms. Ideal for teams building multi-agent systems who need standardized, reusable agent definitions.

Install

Copy this and paste it into Claude Code, Cursor, or any AI assistant:

I want to set up the "Agent Definition File Specification v3.0" 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/klappe-pm/_lattice_lock_framework/main/agents/agents_config/AGENT_SPEC_V3.md"

Then explain what the agent does and how to invoke it.

Description

A governance-first framework for AI-assisted software development with intelligent model orchestration.

Overview

This specification defines a structured format for creating agent and sub-agent definition files that are optimized for Large Language Model (LLM) consumption. These files enable consistent, reliable, and interoperable agent behavior across different LLM platforms and execution environments.

Capabilities Definition

Capabilities explicitly define what the agent can and should do: `yaml capabilities: core_capabilities: • id: "diagnose_issues" description: "Analyze and diagnose technical problems" required_inputs: • name: "error_description" type: "string" required: true • name: "system_context" type: "object" required: false expected_outputs: • name: "diagnosis" type: "string" • name: "confidence_score" type: "number" • name: "next_steps" type: "array" error_handling: "retry_with_backoff" • id: "provide_solutions" description: "Generate step-by-step solutions for identified problems" required_inputs: • name: "problem_diagnosis" type: "string" required: true expected_outputs: • name: "solution_steps" type: "array" • name: "estimated_time" type: "number" • name: "success_probability" type: "number" conditional_capabilities: • id: "escalate_to_human" description: "Escalate complex issues to human support" trigger_conditions: • "confidence_score < 0.6" • "customer_frustration_level > 0.8" • "issue_complexity == 'high'" required_inputs: • name: "conversation_history" type: "array" • name: "issue_summary" type: "string" learning_capabilities: can_learn_from_interactions: true memory_type: "episodic" retention_period: "30_days" update_frequency: "after_each_session" `

Base Format

Agent definition files shall use YAML as the primary format, with JSON as an acceptable alternative. Files must use UTF-8 encoding and follow the naming convention: {domain}_{agent-name}.agent.yaml or {domain}_{agent-name}.agent.json.

Root Structure

`yaml version: "3.0" metadata: id: "uuid-v4-string" name: "human-readable-agent-name" description: "comprehensive-description" author: "author-or-organization" created: "ISO-8601-timestamp" modified: "ISO-8601-timestamp" tags: ["category", "capability", "domain"] agent: role: {} capabilities: {} constraints: {} context: {} behaviors: {} tools: {} sub_agents: [] evaluation: {} adaptation: {} `

Discussion

0/2000
Loading comments...

Health Signals

MaintenanceCommitted 2mo ago
Active
AdoptionUnder 100 stars
0 ★ · Niche
DocsREADME + description
Well-documented

GitHub Signals

Issues5
Updated2mo ago
View on GitHub
MIT License

My Fox Den

Community Rating

Sign in to rate this booster

Works With

Claude Code
Claude.ai