create-agent
Build agents from scratch by generating agent.yaml configuration files, defining agent personality in SOUL.md, and organizing skills, tools, and knowledge directories. Includes templates for model settings, runtime constraints, and sub-agent delegation.
create-agent sets up agent.yaml files, SOUL.md personality definitions, and directory structures for new agents.
AI-generated summary based on this skill's SKILL.md
Decision gist · record as of 2026-07-02
create-agent sets up agent.yaml files, SOUL.md personality definitions, and directory structures for new agents. Build agents from scratch by generating agent.yaml configuration files, defining agent personality in SOUL.md, and organizing skills, tools, and knowledge directories. Includes templates for model settings, runtime constraints, and sub-agent delegation.
Use it when
- create-agent uses SOUL.md to capture your agent's personality, values, and behavioral guidelines.
- create-agent organizes agents with a standard directory layout: place agent.yaml and SOUL.md at the root.
Verify before relying
Read SKILL.md below before installing (1 file). Open directory: indexed for reading, not audited.
Install
open-gitagent/opengap/create-agent · repository language: TypeScript
Open directory. Skills are indexed for reading, not audited. Review a skill's body before installing it.
Frequently asked questions
AI-generated answers based on this skill's SKILL.md and metadata
How do I create an agent yaml file with create-agent?
create-agent helps you generate an agent.yaml configuration file that defines your agent's core settings. The file specifies the agent's model, system prompt, available skills and tools, knowledge base paths, and runtime constraints. You structure your agent directory with agent.yaml at the root, then organize skills, tools, and knowledge subdirectories beneath it. The YAML file acts as the blueprint that the agent runtime reads to initialize and run your agent.
What should I include in SOUL.md to define agent identity?
create-agent uses SOUL.md to capture your agent's personality, values, and behavioral guidelines. This file defines how your agent thinks, communicates, and makes decisions. Include your agent's core identity, communication style, ethical principles, and any domain-specific expertise or constraints. SOUL.md works alongside agent.yaml—while YAML handles technical configuration, SOUL.md provides the narrative context that shapes your agent's responses and reasoning.
How should I set up agent directory structure?
create-agent organizes agents with a standard directory layout: place agent.yaml and SOUL.md at the root, then create subdirectories for skills, tools, and knowledge. Skills contain reusable agent capabilities, tools hold external integrations or utilities, and knowledge stores domain-specific data or documents. This structure keeps your agent modular and maintainable. The agent.yaml file references these directories so the runtime can load and execute them when needed.
What configuration fields and options does agent.yaml support?
create-agent's agent.yaml includes fields for model selection, system prompt, skill and tool declarations, knowledge base paths, and runtime constraints. You can specify which LLM to use, set token limits, define temperature and other sampling parameters, list available skills with their descriptions, declare external tools, and configure sub-agent delegation if needed. The reference documentation shows all available options so you can tailor the agent to your use case.
Can I add a knowledge base and runtime constraints to my agent?
Yes—create-agent supports adding knowledge base directories and runtime constraints to your agent configuration. In agent.yaml, you can point to knowledge directories containing documents, embeddings, or retrieval indexes. Runtime constraints let you set token budgets, timeout limits, and execution policies. These settings ensure your agent operates within defined boundaries while having access to the domain knowledge it needs to answer questions accurately.
How do I configure skills and tools in my agent?
create-agent lets you organize skills and tools in dedicated directories within your agent folder. In agent.yaml, declare each skill with its name, description, and input/output schema so the agent knows when and how to use it. Tools are external integrations—APIs, databases, or utilities—also declared in the YAML with their endpoints and capabilities. The agent runtime loads these definitions and can invoke them during execution based on user requests.
SKILL.md
Rendered from the published skill. Quoted content, verbatim.
Create & Configure Agents
Quick Start
- Create directory structure:
mkdir -p my-agent/skills - Write
agent.yamlwith required fields (see below) - Create
SOUL.mdwith agent identity - Add skills, tools, and knowledge as needed
- Validate:
opengap validate -d ./my-agent
agent.yaml Reference
Required Fields
spec_version: "0.1.0"
name: my-agent # Unique identifier
version: 1.0.0 # Semantic version
description: What this agent does
Model Configuration
model:
preferred: claude-sonnet-4-5-20250929
fallback:
- claude-haiku-4-5-20251001
constraints:
temperature: 0.2 # 0.0 - 1.0
max_tokens: 4096
top_p: 0.9
Skills & Tools
```yaml skills: - code-review # Must
(truncated - see the full file via the links below)
File tree — 1 file
examples/gitagent-helper/skills/create-agent/SKILL.md
Let your AI agent find skills like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 56,283 agent skills by what they can do, searchable in plain language.
wish › “Create and configure a new agent with yaml and personality”
Give your agent the search over MCP, or paste the wish link into any chat. No install? Search from any chat →
Related skills
This skill guides you through installing the gitagent CLI and creating your first agent from a template. You'll customize core files like agent.yaml and SOUL.md, validate your setup, and run your agent locally or from a git repository.
Translate agent configurations across multiple frameworks with a single command. Export to Claude Code, OpenAI, CrewAI, Lyzr, GitHub Models, and other formats, or import from existing Claude, Cursor, and CrewAI projects to standardize your automation setup.
Packet Factory enforces a gatekeeper model where every task with side effects requires an APPROVED packet before running. Create packets in DRAFT status, review them for approval or rejection, then dispatch only approved work. Once approved, scope is locked to prevent creep—expansions spawn new packets instead.
Launch agents with flexible LLM provider selection across Claude, OpenAI, CrewAI, Lyzr, GitHub Models, and other adapters. Supports local directories, remote git repositories, and one-shot prompts for immediate execution. Auto-detection identifies the best adapter from your repo's configuration.
Wiki Ingest automates the process of reading source documents and weaving their content into your wiki. It identifies key entities, claims, and relationships, then creates or updates wiki pages while maintaining bidirectional cross-references and a source index. Each ingest operation is logged for full traceability.
Web Search taps into live web content to surface recent news, articles, and current data on any topic. It returns up to 5 results per query in a structured format with source URLs for easy citation, making it ideal when you need information beyond your existing knowledge base.
More skills paper-search (MIT) · knowledge-retrieval (MIT) · regulatory-analysis (MIT) · lwy-project-agent-writer (MIT) · gmail-email (MIT)