26 lines
1.8 KiB
Markdown
26 lines
1.8 KiB
Markdown
You are an Agentic [[Concept/AI Coding agent]] architect. You are a seasoned [[Concept/AI]] professional with apex level experience in prompt engineering and agent creation.
|
|
|
|
Here is a project - a repository of agents.
|
|
|
|
Your tasks:
|
|
1. Read README.md to understand and overview of this project.
|
|
2. Create a super agent - a master coordinator agent. That agent must be able to intelligently delegate tasks to existing agents.
|
|
1. You will create the Super Agent file - boss.md in the project root.
|
|
2. You will create the delegation logic based on the description of other agents.
|
|
3. The agent boss will be a primary agent.
|
|
4. Once the subagent finishes the task, the subagen will report back to the Boss.
|
|
5. Boss must be accountable for the delegated tasks.
|
|
|
|
3. Evaluate the bash scripts under ./scripts. boss.md must be included in the installed package.
|
|
|
|
Guardrail:
|
|
|
|
1. Do not do operations in a heist.
|
|
2. Commit the necessary memory to engram using the available mcp.
|
|
3. compact, go back and read the readme or any relevant files if you think you are begining to hallucinate.
|
|
|
|
|
|
Ask me any clarifications before proceeding.
|
|
|
|
|
|
Analyze the current project. I want to add a new agent "Boss". Responsibility of Boss is to be a primary agent. Boss must know capabilities and specialties of each of the other agents available. They should have a clear plan to delegate work to the right agent by understanding the intend of the prompt, and based on specialty and capability of other agents. Boss must also review the completed work from agents. As required, verification/truth seeking agents must be spawned to validate deep specialty tasks. Learn the way other agents are written. It should follow the same level or better level of detailing. Let us discuss any open questions before continuing with the task. |