Manus Navigator Activity Sheet
Autonomous Intelligence Lab
Natural Language Agent Systems
Name:
Date:
Period:
Zero-Code Agent Directives
Manus Navigator Guide
Autonomous AI agents execute complex multi-step research, analysis, and problem-solving. Instead of writing Python or JavaScript, you steer the agent using structured natural language logic.
Autonomous Workflow
1. Understanding the Shift: Chatbot vs. Autonomous Agent
Standard Chatbot (Passive)
Responds once to each prompt. It generates text from existing memory but cannot inspect web pages, run tests, or plan iterative steps.
Prompt → Single Text Reply
Manus Autonomous Agent (Active)
Breaks a goal into sequential sub-tasks: searches primary web sources, gathers data, cross-references findings, and produces deliverables.
Goal → Plan → Act → Verify
2. The 4-Part Directive Protocol (Code-Free Architecture)
A. Primary Objective
Define the concrete end goal clearly without ambiguity.
"Investigate 2026 ocean cleanup prototypes..."
B. Guardrails & Rules
Set restrictions, source types, and forbidden shortcuts.
"Only peer-reviewed studies, max 3 years old..."
C. Execution Steps
Provide logical sequencing milestones for the agent.
"1. Find 3 designs 2. Compare cost 3. Identify flaws"
D. Deliverable Spec
Specify exact format, tables, citations, or summary structure.
"A comparison matrix table with cited URLs."
3. Classroom Agent Missions (Select One for Today's Lab)
MISSION 1
The Clean Energy Feasibility Audit: Instruct Manus to evaluate floating solar farms vs. offshore wind in coastal cities, compiling installation costs, energy output per sq km, and environmental tradeoffs.
MISSION 2
The Urban Heat Island Countermeasure: Command Manus to research 3 world cities using green roofs and reflective pavements, synthesizing measurable temperature drops and municipal policy frameworks.
MISSION 3
Deep Space Mission Risk Analyzer: Direct Manus to synthesize cosmic radiation mitigation strategies for a crewed Mars journey, contrasting magnetic shielding against water-wall passive absorption.
Turn to Page 2 to draft your directive, monitor execution, and audit agent findings. Page 1 of 2
Mission Execution & Human Oversight Log
Selected Mission:
1 Draft Your Natural Language Directive (No-Code Prompt) Write in complete instructions
Objective:
Constraints & Prohibited Shortcuts:
Step Sequence & Required Output Format:
2 Live Flight Log: Track the Agent's Autonomous Actions
As Manus operates, watch its browser tabs, search terms, and intermediate calculations. Log three distinct actions it took on its own.
| Sub-Task | Action / Tool Used (Search, Visit URL, Code) | Primary Observation or Data Found |
|---|
| Step 1 | | |
| Step 2 | | |
| Step 3 | | |
3 Human-in-the-Loop Audit: Fact-Check & Verification
Verification Checklist:
Did the agent cite live, valid source URLs?
Are data numbers consistent across tables?
Did it respect all stated constraints?
One Fact Checked with Direct Source:
4 Directive Iteration: How Would You Refine the Prompt?
Where did the autonomous agent get slightly off track or take an inefficient detour? What rule would you add next time?
Manus AI Classroom Investigation • Autonomous Agent Literacy Page 2 of 2
