oracle

Bundle prompts with selected repository files into one-shot AI requests.

Updated Apr 8, 2026
One-click install
npx skills add https://github.com/JOHNNYWHITEMIKE/openclaw --skill oracle-johnnywhitemike
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: oracle
Source: https://github.com/JOHNNYWHITEMIKE/openclaw/tree/main/openclaw/.openclaw/openclaw/skills/oracle
Command: npx skills add https://github.com/JOHNNYWHITEMIKE/openclaw --skill oracle-johnnywhitemike

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Oracle eliminates the friction of giving an AI the right repository context by bundling your prompt together with a precise set of files into a one-shot request so the model can answer with grounded information.

Core Features & Use Cases

  • Prompt + file bundling: Attach a curated set of source files (including or excluding globs) to ensure the model sees the truth rather than guesses.
  • Dry-run and file reporting: Preview payload scope and token impact before you spend tokens.
  • Session-based browser runs: Use long-running browser mode with reattachable stored sessions to avoid losing progress.
  • Common use cases: Codebase Q&A, debugging with verbatim error context, refactoring plans, and “restore context later” investigations.

Quick Start

Ask the oracle skill to diagnose the bug in your codebase by running in browser engine with a focused file set: oracle --engine browser --model gpt-5.2-pro -p "Find the root cause of this failing test and propose a minimal fix with a patch plan and any needed test updates." --file "src/" --file "!/.test.".

Frequently Asked Questions about oracle

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

FAQPage Schema
How do I bundle local codebase files into an AI prompt for grounded debugging?

You can bundle local codebase files into an AI prompt by attaching a curated set of source files using include and exclude file globs. This ensures the model answers using real repository context rather than assumptions.

Can I preview token usage and file scope before sending a large repository context to the model?

Yes, you can preview token usage and file scope using a dry-run preview. This feature reports the payload scope and calculates token impact before you spend tokens on the actual request.

How does prompt bundling work with browser sessions for long-running codebase Q&A?

Prompt bundling works with browser sessions by using long-running browser mode with reattachable stored sessions. This allows you to avoid losing progress during extended codebase Q&A or debugging workflows.

What is the best way to include specific source files while excluding tests in an AI refactoring plan?

The best way to include specific source files while excluding tests is by applying file glob patterns. You use include patterns for source directories and exclude patterns like "!**/*.test.*" to filter out test files.

Do I need YAML-defined discovery metadata to use file globs for repository-grounded AI help?

Yes, you need YAML-defined discovery metadata to use file globs for repository-grounded AI help. This metadata configuration supports the include and exclude file glob patterns required for precise file selection.

Can I select different engines for browser or API usage when asking an AI to find a bug?

Yes, you can select different engines for browser or API usage when asking an AI to find a bug. The skill supports engine selection to run requests in either browser mode or via API.