oracle

Bundle prompts and repository files for browser-based ChatGPT consultation.

13|3|Updated Mar 2, 2026

Search Tags:#debugging #context #oracle #prompts #session #browser #code-review

Ready to run

npx skills add https://github.com/misty-step/harness-kit --skill oracle-misty-step

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: oracle
Source: https://github.com/misty-step/harness-kit/tree/main/skills/oracle
Command: npx skills add https://github.com/misty-step/harness-kit --skill oracle-misty-step

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Browser-mode-only Oracle consults: bundle a prompt plus selected files and ask a signed-in ChatGPT browser session for a second opinion. Use when stuck, debugging hard bugs, reviewing an architecture plan, or cross-checking a substantive diff with large file context. Never use Oracle API mode from Harness Kit. Trigger: /oracle, /consult.

Core Features & Use Cases

Browser-backed consultation: Bundle prompts and files and request expert review from a signed-in browser session.
Contextual analysis: Compare changes against large code or design diffs with preserved file context.
Synthesis and guidance: Lead agent synthesizes, verifies, and delivers final judgment while keeping user data safe.

Quick Start

Trigger a browser-backed consult using the /oracle trigger with your task prompt and the relevant file globs.

Frequently Asked Questions about oracle

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I get a second opinion on a substantive code diff with large context?▼

You can get a second opinion on a large code diff by bundling your prompt and selected repository files to consult a signed-in ChatGPT browser session for expert review and synthesized guidance.

Can I use my signed-in browser session for debugging hard bugs?▼

Yes, you can use your signed-in ChatGPT browser session for debugging hard bugs by bundling the prompt with relevant files to receive contextual analysis and expert verification.

How do I review an architecture plan against large file context?▼

Review an architecture plan against large file context by bundling your design documents and relevant code globs into a prompt, then consulting a browser-backed session for contextual analysis and synthesis.

Does the Oracle Skill require API mode or OpenAI providers?▼

No, the Oracle Skill avoids API mode and OpenAI providers entirely; it enforces running with the browser engine and prefers the gpt-5.5-pro model for all consultations.

What is the best way to trigger a browser-backed consultation for code review?▼

The best way to trigger a browser-backed consultation is using the /oracle or /consult command with your task prompt and relevant file globs to request expert review from a signed-in browser session.

Where are consultation sessions stored when running a browser-backed code review?▼

Consultation sessions from a browser-backed code review are stored under the ~/.oracle/sessions directory to preserve context and history for future debugging and architecture analysis.