oracle

Delegates complex debugging, security audits, and architecture reviews to GPT-5 CLI tools.

3|Updated Jan 15, 2026
One-click install
npx skills add https://github.com/trudyan141/my-antigravity-agents-kit --skill oracle-trudyan141
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: oracle
Source: https://github.com/trudyan141/my-antigravity-agents-kit/tree/main/templates/.agent/skills/oracle
Command: npx skills add https://github.com/trudyan141/my-antigravity-agents-kit --skill oracle-trudyan141

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex bugs, architectural decisions, and security reviews often exceed what a single-pass analysis can catch. This Skill routes difficult problems to GPT-5 via available CLI tools (cursor-agent, codex, or opencode) for deep second opinions, and falls back to structured in-house analysis frameworks when those tools are unavailable. ## Core Features & Use Cases - GPT-5 CLI Delegation: Automatically detects cursor-agent, codex, or opencode and forwards the user's complete request to GPT-5 with no modification. - Advanced Debugging Playbooks: Built-in methodologies for race conditions, memory leaks, performance bottlenecks, and concurrency issues. - Security & Architecture Audits: Checklists for authentication review, input validation, cryptographic usage, coupling/cohesion analysis, and scalability assessment. - Use Case: You hit an intermittent race condition in async TypeScript code. Invoke the Oracle to get a systematic root-cause analysis with a severity-rated report, mitigation strategies, and validation steps. ## Quick Start Ask the Oracle to review the recent commit for correctness and security issues, or to analyze a complex bug you cannot reproduce consistently.

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 complex bug using GPT-5?

Invoke the Oracle with your complete problem description, and it forwards the request to GPT-5 via cursor-agent, codex, or opencode depending on which CLI is installed. If none are available, it applies its own systematic debugging frameworks instead.

How to review a git commit for correctness and security issues?

The Oracle analyzes commit scope using git diff, then reviews logic correctness, edge cases, performance implications, security considerations, and backward compatibility. Findings are reported with severity ratings and specific code-level fixes.

What CLI tools does the Oracle support for GPT-5 access?

It checks for cursor-agent first, then codex, then opencode, in that priority order. Each is invoked with the GPT-5 model flag, and the raw GPT-5 response is returned unmodified.

What happens if cursor-agent, codex, and opencode are not installed?

The Oracle falls back to its own analysis capabilities using built-in frameworks for debugging, security auditing, and architecture review. It still produces structured reports with findings, action items, and validation steps.

Can the Oracle make code changes directly?

No, the Oracle only provides analysis, insights, and recommendations. Its role is to illuminate problems and guide solutions, not to modify code, and it is restricted to the Bash tool for CLI delegation.