The library

Every prompt, annotated

Real, published system prompts from shipping AI products — each with source attribution and notes on what makes it work.

100 prompts
LE
Leap.new
system prompt

Leap.new — Encore.ts Full-Stack App Builder System Prompt

System prompt for Leap.new, a prompt-to-app builder that generates full-stack applications with an Encore.ts backend and a React/Vite/Tailwind/shadcn …

role definition custom XML output schema constraint setting
31 7.3k 5.2k words
MA
Manus
agent prompt

Manus — Agent Loop Core Prompt

Condensed core prompt for the Manus autonomous agent describing its identity, task strengths (research, data analysis, content/website creation, codin…

role definition agent loop specification capability enumeration
31 8.7k 360 words
Trae
tool schema

Trae — Builder Mode Tool Schemas

JSON tool-schema definitions for Trae's agentic Builder mode, covering 18 tools spanning task tracking, codebase retrieval, file editing, command exec…

tool-use protocol output formatting constraint setting
23 1.8k 740 words
VSCode Agent (GitHub Copilot)
agent prompt

VSCode Agent — GitHub Copilot Agent Prompt with Tools

Full agent system prompt for GitHub Copilot in VS Code, combining an identity/instructions block with an embedded JSON array of ~20 tool definitions a…

role definition XML sectioning tool-use protocol
21 5.3k 2.6k words
Lumo
system prompt

Proton Lumo — Privacy Assistant System Prompt

System prompt for Proton's privacy-focused chat assistant Lumo. It sets an identity and personality, web-search triggering criteria gated behind a use…

role definition persona tool-use protocol
16 5.4k 1.3k words
RO
RooCode
agent prompt

RooCode — Code Mode System Prompt

Full system prompt for RooCode's 'Code' mode, a VS Code agent. It opens with UI/configuration text, then defines an XML-tag tool-use protocol with one…

role definition tool-use protocol XML output formatting
13 4.3k 7.2k words
DE
Devin
agent prompt

Devin AI — Autonomous Software Engineer Prompt

The full system prompt for Devin, an autonomous software engineering agent operating a real OS. It defines a planning vs standard mode model, a <think…

role definition tool-use protocol chain-of-thought
8 5.3k 4.9k words
MA
Manus
system prompt

Manus — Full Modules System Prompt

The full tagged system prompt for the Manus agent, structured into many XML-style rule modules: intro, language_settings, system_capability, event_str…

modular XML-tagged sections role definition event-stream protocol
8 7.9k 1.7k words
PO
Poke
agent prompt

Poke — Execution Engine Agent Prompt

System prompt for Poke's backend 'execution engine' agent, which serves a separate user-facing Poke personality rather than the user directly. It defi…

role definition tool-use protocol few-shot
4 4.7k 1.9k words
LO
Lovable
tool schema

Lovable — Agent Tool Schema Definitions (JSON)

JSON object defining Lovable's agent toolset for editing React projects, including file operations (lov-write, lov-line-replace, lov-view, lov-rename,…

tool description as embedded protocol few-shot examples in descriptions constraint setting
3 480 3.3k words