Junie — Read-Only Exploration Agent

Agent Prompt ✦ Curated & attributed Intermediate

System prompt for Junie operating in a strict read-only investigation mode. It defines a small set of project-exploration commands (search_project, get_file_structure, open, open_entire_file, goto, scroll_down/up, answer) plus standard read-only bash, and enforces a one-command-per-turn loop where each response is wrapped in <THOUGHT> and <COMMAND> XML tags and the session terminates only via the answer command.

What makes this prompt notable

  • Hard-enforces read-only mode: 'don't modify, create or remove any files'
  • Requires every response to contain exactly one <THOUGHT> and one <COMMAND> tag, with nothing outside them, and one command at a time
  • Documents the fuzzy search_project tool's scoping behavior (e.g. 'class User' narrows to the definition vs bare 'User' for exhaustive matches)
  • Routes general non-project questions straight to the answer command without exploration

Discussion (0)

Sign in to join the discussion.
Related

More in IDE & Coding Assistants

VSCode Agent
agent prompt

VSCode Agent (GitHub Copilot) — GPT-5 Agent Prompt

The most elaborate Copilot VS Code agent prompt, for GPT-5. It combines the Copilot identity and refusals with a detailed communication-style guide, m…

role definition tool-use protocol structured diff format specification
239 5.6k 3.7k words
KI
Kiro
system prompt

Kiro — Intent Mode Classifier

A lightweight intent-classification prompt that routes a user message into chat/do/spec modes and returns a JSON object of confidence scores summing t…

classification rubric constraint setting few-shot
226 5.3k 600 words
CU
Cursor
agent prompt

Cursor — Agent Prompt v1.0 (Claude Sonnet 4)

An earlier, leaner Cursor agent prompt (Claude Sonnet 4) focused on tool-calling rules, aggressive parallel tool execution, search-and-reading bias to…

role definition tool-use protocol parallel tool-call maximization
225 7.3k 1.5k words