System prompts
Inside the machine
Educational catalog of how coding agents, chat apps, and research tools instruct models — token weight, detected techniques, and one-click weigh. Updated 2026-09-26.
Showing 39 of 39 catalog prompts · Compare two →
Cursor-style agent
Autonomous IDE agent pattern: tools, edits, and terse engineering tone.
7/9 techniques detected
CLI coding agent
Terse tool-driven CLI agent for repo tasks.
6/9 techniques detected
Generative UI builder
React/Next UI generator with component constraints.
5/9 techniques detected
In-browser IDE agent
WebContainer-style full-stack agent for browser IDEs.
5/9 techniques detected
General chat assistant
Default helpful chat posture with safety and formatting.
4/9 techniques detected
Cascade-style IDE agent
Memory-aware batched edits for agentic IDEs.
4/9 techniques detected
Cloud IDE agent
Search-first, workflow-based autonomous programmer.
5/9 techniques detected
Research browser assistant
Citation-oriented research assistant with browsing tools.
3/9 techniques detected
Workspace writing assistant
Docs/productivity assistant for structured writing.
3/9 techniques detected
App builder agent
Realtime React app builder with live preview constraints.
1/9 techniques detected
Pair programmer chat
Inline IDE pair-programmer for explanations and patches.
3/9 techniques detected
Autonomous engineer planner
Plan/standard mode workflow for long-running coding agents.
5/9 techniques detected
Augment-style code agent
GPT-oriented coding agent with repo grounding and patch discipline.
3/9 techniques detected
Same.dev design-to-code
Cloud IDE focused on design-to-code with Bun/Next preferences.
3/9 techniques detected
Trae Builder mode
ByteDance Trae-style IDE builder agent.
6/9 techniques detected
RooCode agent
Open-source RooCode-style IDE agent with strong tool use.
2/9 techniques detected
Kiro Spec mode
AWS developer IDE — Spec mode for structured implementation.
4/9 techniques detected
Junie IDE agent
JetBrains Junie-style coding assistant.
3/9 techniques detected
Warp terminal agent
AI terminal assistant for shell workflows.
3/9 techniques detected
Dia browser assistant
Browser-company assistant for page-aware chat.
5/9 techniques detected
Manus research agent
Planner + knowledge modules for general research agents.
2/9 techniques detected
Microsoft Copilot
Enterprise assistant posture with grounding and safety.
3/9 techniques detected
Grok assistant
xAI Grok-style helpful chat with wit dialed carefully.
3/9 techniques detected
Gemini assistant
Google Gemini-style multimodal-ready assistant prompt.
4/9 techniques detected
VS Code Agent
Microsoft VS Code agent for autonomous editing.
2/9 techniques detected
Xcode assistant
Apple Xcode AI coding assistant patterns.
3/9 techniques detected
Orchids coding agent
Orchids.app-style coding agent for web apps.
2/9 techniques detected
Qoder coding agent
Qoder-style software engineering agent.
2/9 techniques detected
Z.ai Code agent
Z.ai coding agent patterns.
3/9 techniques detected
Aider pair programmer
Git-aware pair programmer for repo edits and commit messages.
5/9 techniques detected
Cline agent
VS Code agentic loop with plan, tools, and user checkpoints.
6/9 techniques detected
Continue.dev assistant
Open IDE assistant for chat, edit, and codebase Q&A.
4/9 techniques detected
Amazon Q Developer
AWS-oriented coding assistant for apps and cloud workflows.
5/9 techniques detected
Sourcegraph Cody
Codebase-aware chat for large repos and code search.
5/9 techniques detected
OpenHands agent
Open-source software agent with shell, browser, and editor tools.
4/9 techniques detected
SWE-agent
Issue-solving agent loop for GitHub-style software tasks.
2/9 techniques detected
Phind engineer
Developer Q&A assistant optimized for technical search answers.
2/9 techniques detected
Tabnine chat
Privacy-minded code completion chat for teams.
3/9 techniques detected
JetBrains AI Assistant
IDE assistant for explain, refactor, and generate inside JetBrains IDEs.
3/9 techniques detected
What is a system prompt?
The invisible instruction layer that turns a generic LLM into a product. Read the guides or weigh any prompt on the token counter.