name: permanent-agent-creator description: Guide for creating and updating permanent agents with stable names, dedicated system prompts, and optional workspace folders. license: Complete terms in LICENSE.txt

Permanent Agent Creator

This skill provides guidance for creating and updating permanent agents.

About Permanent Agents

Permanent agents are background agents with stable identities. They are created once and reused across sessions, with a dedicated system prompt that can be updated over time.

What Permanent Agents Provide

1. Stable identity - predictable agent ids you can reference later
2. Dedicated system prompts - durable instructions for specialized roles
3. Optional workspace folders - focused file scopes inside the main workspace
4. Reusable collaboration - multiple agents can coordinate through a named helper

Core Principles

Keep Prompts Focused

Permanent agents should be narrow and durable. Prefer a short system prompt that captures role, constraints, and communication style. Avoid task-specific instructions that should live in the user request instead.

Stable Names Matter

Names are how humans recognize agents. Pick a short, descriptive name and avoid frequent renames. If a name needs to change, update it deliberately alongside the system prompt.

Scope Workspace When Needed

Only assign a workspace subfolder when the agent needs a dedicated area (e.g., a specific project or dataset). Default to the main workspace when no isolation is required.

Workflow

1. Gather inputs

   • Agent name
   • Short description
   • System prompt
   • Optional workspace subfolder

2. Create or update Use create_permanent_agent with the collected inputs. Reusing a name updates the existing agent.

3. Share the agent id Store the returned agent id and use send_agent_message to coordinate with it.

Example

Create a release helper:

create_permanent_agent({
  name: "Release Tracker",
  description: "Tracks release readiness and reports risks.",
  systemPrompt: "You track release status, summarize risks, and report progress to the main agent.",
  workspaceDir: "release-notes"
})

Then delegate work:

send_agent_message({
  agentId: "<agent-id>",
  text: "Summarize the latest release checklist status."
})