Skip to content

Oracle CLI Best Practices

Bundle a prompt and selected project files into one request for an outside model via the oracle CLI

Stars
11,367
Licence
MIT
Files
1 file

What it does

This skill teaches an agent how to use the oracle command-line tool, which packages a question plus chosen project files into a single one-shot request that another AI model (such as GPT-5.2 Pro through ChatGPT in a browser, or through an API) can answer with real repository context. It covers previewing what will be sent, attaching the right files with include and exclude patterns, reattaching to long-running sessions instead of re-running them, and writing high-signal prompts. The agent treats the returned answer as advice to verify against code and tests.

Great for

  • Developers who want a second model to review a tricky bug or design decision using their actual code
  • Teams that already use the oracle CLI and want consistent file selection and prompt habits
  • People comfortable with long-running browser-based model sessions that can take 10 minutes to an hour

Skip it if

  • Users who need a fast answer: the default browser workflow is described as often taking 10 minutes to an hour, which is slow for quick questions
  • Projects with secrets in the files being attached: the skill advises redacting, but any attached file is sent to a third-party model service, so sensitive code or credentials could be exposed if care is not taken
  • Users without the oracle CLI installed or without access to ChatGPT or an OpenAI API key, since the skill depends on those services

Try asking

  • “Preview what files would be sent for a question about my src folder before spending any tokens”
  • “Ask GPT-5.2 Pro in the browser to review my authentication module and suggest a patch plan with tests”
  • “Reattach to my stored oracle session from yesterday instead of re-running a long analysis”
  • “Write a standalone prompt and file set for a multi-day investigation so I can rerun it later”
Step-by-step guide: how to use Oracle CLI Best PracticesInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    oracle.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip oracle.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip oracle.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run oracle --dry-run summary --files-report first to check the file set and token size before a real run
  • Exclude test files, snapshots, and secrets with negative patterns such as !**/*.test.* and !**/*.snap, and never attach .env or key files
  • When a browser run times out, reattach with oracle session <id> rather than starting over, since re-running wastes time and may hit the duplicate-prompt guard

What's inside

  • SKILL.md

Redistributed unmodified from spinabot/brigade under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.