Evidence-first authority page
How to Write Effective System Prompts for AI Agents
Practical guidance for writing agent system prompts that produce reliable tool-use behavior, beyond general prompt engineering advice.
Direct answer
An effective agent system prompt clearly defines the agent's role, explicitly states when to use each available tool and when not to, specifies the stopping condition, and includes instructions for handling tool failures — vague or missing guidance on any of these four areas is a common root cause of unreliable agent behavior.
Last reviewed 2026-08-29
This page remains noindex while its page-level source receipts are completed and verified. It is available for review, not presented as publication-ready authority.
The four things every agent system prompt needs
- Role definition — what the agent is and isn't responsible for.
- Tool usage guidance — specifically when to call each tool, not just what each tool does.
- Stopping condition — how the agent knows the task is complete.
- Failure handling — what to do when a tool call errors or returns unexpected results, rather than leaving this undefined.
Test against edge cases, not just the happy path
A system prompt that produces correct behavior on straightforward inputs can still fail on ambiguous or edge-case inputs — test specifically against cases where the right tool choice or stopping point isn't obvious, since that's where prompt ambiguity actually surfaces.