oracle

Bundle prompts and repository files into a single context payload.

Updated Mar 13, 2026
One-click install
npx skills add https://github.com/Yogi-076/Vajrascan.on --skill oracle-yogi-076
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: oracle
Source: https://github.com/Yogi-076/Vajrascan.on/tree/main/moltbot/skills/oracle
Command: npx skills add https://github.com/Yogi-076/Vajrascan.on --skill oracle-yogi-076

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Oracle removes the guesswork of providing repository context to large models by bundling a prompt with a curated set of files so a model can reason about real project code and configs in a single one-shot request.

Core Features & Use Cases

  • Context bundling: Select files, directories, and globs to include or exclude so the model gets only the most relevant project context.
  • Engine and session control: Choose browser or API engines, preview token spend with dry-runs, and reattach to long-running stored sessions.
  • Safety and ergonomics: Honors .gitignore, excludes large files and common build dirs by default, and recommends redacting secrets before attachment.

Quick Start

Run oracle with a tight file set and the browser engine to attach project files and request a focused code-level analysis.

Frequently Asked Questions about oracle

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

FAQPage Schema
How do I bundle repository files into a single prompt for a one-shot LLM run?

You can bundle repository files into a single context payload by selecting specific files, directories, and globs to include or exclude, ensuring the model receives only the most relevant project context for a one-shot request.

Can I preview token spend before sending files to the model?

Yes, you can perform token-preview dry-runs to estimate token spend before executing the actual model run, allowing you to adjust file inclusion patterns and manage context window limits.

Does the context bundling process automatically ignore node_modules and build directories?

The context bundling process honors .gitignore rules, excludes large files and common build directories like node_modules by default, and recommends redacting secrets to maintain safety constraints.

How do I reattach to a long-running browser or API LLM session?

You can reattach to long-running stored LLM sessions by utilizing the session management features, which support both browser and API engine selection while preserving the original context payload.

What is the best way to provide real project code to a large model for code-level analysis?

The best way is to use a tight file set with globbing patterns to curate relevant project code and configs, bundling them with your prompt for a focused, context-aware one-shot model run.

When should I use the browser engine versus the API engine for long-running model sessions?

You should select the browser or API engine based on your workflow requirements, utilizing the browser engine for browser-based LLM interactions and the API engine for direct API sessions.