agent-system-prompt
Design and generate system prompts for AI agents, assistants, and chatbots. Use when creating, writing, or improving a system prompt, persona, role definition, or operating instructions for any AI agent (assistant, coding, autonomous, research, support, or orchestrator).
Agent System Prompt
Generate a complete system prompt for an AI agent from plain-language requirements, using a reusable component library and a deterministic assembly workflow.
Workflow
- Gather requirements — purpose, tools, audience/language, tone, hard constraints, output format, environment. Ask only if a missing item would change the result.
- Select archetype — match the purpose in references/agent-types.md; it lists required/optional components and gives the skeleton.
- Assemble components — fill each required component from references/anatomy.md; include only what the agent will actually use.
- Write — start from the matching template in assets/templates/; follow references/output-patterns.md.
- Validate — run the checklist below.
Quality Checklist
- Identity unambiguous — "You are X" plus what it is and is not.
- One clear mission stated up front.
- Every tool described — what it does and when to use it.
- Tone defined — phrasing to use and to avoid.
- Boundaries explicit — what to refuse, disclaim, or never disclose.
- Output format specified — structure, length, citation conventions.
- Environment/context stated — what's available each turn.
- No dead weight — every section maps to a real capability or constraint.
- Consistent voice — imperative, present tense, second person.
Resources
- references/anatomy.md — component library (building blocks + skeletons).
- references/agent-types.md — archetypes and their required/optional components.
- references/output-patterns.md — writing style, section order, validation.
- assets/templates/ — ready-to-fill templates per archetype.
Related skills
- Downloads
- Security audit
- Review
- Last updated
- 1d ago
- Current version
- v1.0.0
- License
- MIT-0
