Automates hiring specialist agents with permanent contracts for task execution.
npx skills add https://github.com/jota-batuta/batuta-dots --skill agent-hiring
Please help me install this Agent Skill. Skill: agent-hiring Source: https://github.com/jota-batuta/batuta-dots/tree/main/BatutaClaude/skills/agent-hiring Command: npx skills add https://github.com/jota-batuta/batuta-dots --skill agent-hiring
The main agent never executes tasks directly. For every task, it "hires" a specialist agent. Agents are .md files in .claude/agents/ -- PERMANENT contracts that persist across projects. This skill defines the contract-based hiring system.
When a task arises, trigger the agent-hiring workflow to select or create a specialized contract and assign explicit file ownership.
High-intent search queries and answers about installing and using this skill.
To create persistent agent contracts, you write a dedicated .md file under the .claude/agents/ directory with frontmatter defining the name, description, tools, model, skills, and maxTurns. This contract persists across projects and is invoked when the main agent hires a specialist.
The agent hiring workflow executes tasks by having the main agent analyze the task, search local and global agent libraries for existing specialists, and require explicit user approval before creating a new contract file and spawning the hired agent with defined file ownership.
A Claude agent frontmatter contract requires defining the agent's name, description, tools, model, skills, and maxTurns. It must also enforce explicit file ownership guidelines to scope the hired agent's execution permissions.
Yes, specialist agents are saved as .md files in .claude/agents/ and function as permanent contracts that persist across projects. The workflow searches both project-local and global agent libraries to verify existing skills before proposing a new hire.
The main agent does not execute tasks directly because the system uses a contract-based hiring model where every task is delegated to a specialist agent. This requires explicit user approval and file ownership scoping to ensure controlled, specialized task execution.
The best way to scale agent automation is using permanent .md contract files in .claude/agents/. The system searches local and global libraries, verifies skills are current, and prompts for approval before creating new contracts, ensuring consistent specialist invocation across projects.
UsernameTronDefine a complete YAML frontmatter schema for Claude Code subagent .md files.
involvexGenerate file-based delegation instructions for sub-agent code tasks.
AcosmiManage command execution approvals and allowlists for Crab Claw deployments.
joonlabDevelop LangChain and LangGraph agents with planning, memory, and delegation middleware.
