Back to Glossary

Agent

الوكيل (AI Agent)

GLS-000161

Advancedai-prompting1 min read
ai agentautonomous agentllm agentagentic aimulti-step ai
Official DocsSign in to save & take notes

Definition

An AI system that takes multi-step actions autonomously to complete a goal — using tools and making decisions.

نظام AI يتخذ إجراءات متعددة الخطوات بشكل مستقل لتحقيق هدف — باستخدام الأدوات واتخاذ القرارات.

Why It Matters

Understanding agents helps 404Fault's users understand what AI tools like Claude Code, Cursor, and GitHub Copilot are doing under the hood.

فهم الوكلاء يساعد مستخدمي 404Fault على فهم ما تفعله أدوات AI مثل Claude Code وCursor وGitHub Copilot تحت الغطاء.

Full Definition

An AI agent is an LLM that can use tools (search the web, write files, call APIs, run code) in a loop to complete complex tasks. Unlike a single prompt-response, an agent plans, acts, observes results, and iterates. Claude Code (the tool writing this code) is an AI agent. Agents raise new safety questions: what actions should an autonomous AI be allowed to take?
وكيل AI هو LLM يمكنه استخدام الأدوات (البحث على الويب وكتابة الملفات واستدعاء APIs وتشغيل الكود) في حلقة لإتمام مهام معقدة. على عكس استجابة مطالبة واحدة، يخطّط الوكيل ويتصرف ويلاحظ النتائج ويُكرِّر. Claude Code (الأداة التي تكتب هذا الكود) هو وكيل AI. تطرح الوكلاء أسئلة أمان جديدة: ما الإجراءات التي يجب السماح لـ AI المستقل باتخاذها؟
Knowledge Graph

AI Builder Tips

Sign in for more features

Avoid these mistakes when using Agent:

1

Giving agents too broad permissions (they can take destructive actions if not constrained)

2

Not reviewing agent outputs before applying them to production systems

Sign in to unlock guided AI explanations from AI Teacher.

Generate a Prompt

Copy this prompt and use it directly with any AI model — no setup needed.

Ready-to-Use Prompt
Help me build a project using Agent.

Explain:
1. What is Agent and why it matters
2. The core architecture and required tools
3. Step-by-step implementation plan
4. Common mistakes to avoid: Giving agents too broad permissions (they can take destructive actions if not constrained), Not reviewing agent outputs before applying them to production systems
5. Best practices and production tips

Official Resources

Official Documentation
Browse all AI termsCreate free account