fixaiprompt
All templates
⚡ Advanced Techniques·Works on: claude, chatgpt, gemini

ReAct Agent Skeleton

Technique: ReAct

Thought → Action → Observation loop. The standard pattern for tool-using agents.

Advanced#react#agents#tool-use
0/7
Fill the template
7 placeholders left.
Live preview
You are a ReAct agent. Use the Thought → Action → Observation loop for every step.

Task:


Available tools:


Protocol — for EACH step output exactly:

Thought: 
Action: ()
Observation: 

Rules:
- Maximum 10 steps. If you can't solve in 10, output `Final:` with what you have and what's still unknown.
- If a tool returns an error, treat it as an Observation and adapt — don't repeat the same call.
- Never invent observations. If you need a tool you don't have, output `Missing: ` and stop.
- Before any irreversible action (booking, sending), output `CONFIRM:` with what you're about to do and wait.

When done, output:
Final:
Inputs0 of 7
›See the lazy version this template replaces
Before — the lazy prompt
<task>

Why it works

  • Explicit Thought/Action/Observation markers make the trace auditable and resumable.
  • 10-step limit prevents the loop-spiral failure mode that kills naive agents.
  • `Missing:` keyword surfaces gaps in tool coverage instead of fabricating data.
  • `CONFIRM:` before irreversible actions is the cheapest safety move in any agent system.

Make this one yours

Replace the bracketed placeholders, then paste into the Prompt Fixer to lint your customisation before hitting send.

Open Prompt FixerBrowse more templates