AI CLI Tools

Build copy-ready Claude Code, Codex, and OpenCode commands with shareable preset workflows, a privacy-safe URL, and a clearer workflow summary.

Command builder
claude
Safe share URL
Prompt privacy Custom prompt text stays local and is not added to the share URL.
Workflow Summary
Tool Claude Code
Mode Interactive
Safety posture Default Claude Code permission posture.
Output style Interactive terminal session.
Share mode Current page
Prompt source Private prompt
Compare with Switch to Print Mode when you need one copy-ready answer instead of a live session.
Representative workflows
Claude Code
Claude review loop
Preset
Stay in an interactive Claude Code session when you expect a few follow-up questions before applying the next change.
Compare with Print Mode when you need a single copy-ready answer for scripting or CI.
Claude Code
Claude auto mode run
Preset
Start Claude with auto mode and the auto permission posture for hands-off bulk edits.
Compare with the review loop when you still want to gate every change.
Claude Code
Claude auto mode defaults
Preset
Print the default allow and block rules for Claude auto mode before you trust it on real changes.
Compare with the auto mode run preset when you are ready to actually execute a task.
Claude Code
Claude one-shot handoff
Preset
Use Claude print mode when you want one response for a handoff, summary, or automation script.
Compare with the interactive review loop when the task will need multiple follow-ups.
Codex
Codex exec review
Preset
Use `codex exec` for a repeatable non-interactive pass with an explicit sandbox and approval posture.
Compare with interactive Codex when you want to explore the repo before locking the command for automation.
Codex
Codex interactive exploration
Preset
Start in interactive Codex when the repo is unfamiliar and you want to inspect the codebase before picking an exec command.
Compare with `codex exec` when the task is stable enough for a one-shot scripted run.
OpenCode
OpenCode run with JSON events
Preset
Use `opencode run` when you need a one-shot task run and machine-readable output.
Compare with `opencode serve` when you want to keep a backend warm for repeated runs.
OpenCode
OpenCode warm server
Preset
Run a reusable OpenCode backend when you want repeated requests without paying startup cost each time.
Compare with `opencode run` for one-off commands that do not need a long-lived server.
Frequently used prompts
Review a pull request
Catch regressions, risky behavior changes, and missing tests before human review.
Review this change for regressions, risky behavior changes, missing tests, and documentation gaps. Focus on the highest-signal findings first. Cite the specific files or code paths involved and recommend the next checks to run.
Understand a codebase
Map the request flow, module ownership, and risky files before editing an unfamiliar area.
Explain how the request flows through <feature or system area> in this codebase. Include which modules own what, where validation and side effects happen, and the main gotchas before editing. End with the files I should read next.
Iterate on a difficult problem
Run a measured improvement loop with scoring, artifacts, and explicit iteration notes.
Treat this as an eval-driven improvement loop. Before changing anything, read AGENTS.md and find the command or script that measures success. Make one focused improvement at a time, rerun the checks after each meaningful change, log what improved or regressed, and keep iterating until the quality bar is met.
Upgrade an API integration
Inventory the current integration, migrate carefully, and surface any prompt or response-shape risks.
Upgrade this integration to the latest recommended API and model path. Start by inventorying the current endpoints, models, prompts, and tool assumptions. Choose the smallest migration that preserves behavior, update prompts where the new guidance requires it, and call out any response-shape or manual review risks.
Build from screenshots or notes
Turn references into responsive UI while staying inside the repo's design system and code patterns.
Implement this UI in the current project using the screenshots, mocks, or notes I provide as the source of truth. Reuse the existing design system and component patterns, match the hierarchy and responsive behavior closely, and note any assumptions when a detail is ambiguous. Finish by checking the result against the references.
Kick off a task from a thread
Turn a thread or issue into a scoped implementation plan and a verified end-to-end change.
Analyze the issue or thread I provide and implement the fix or feature in this workspace. Start by summarizing the scope, constraints, risky files, and verification plan. Then make the smallest end-to-end change that satisfies the request and finish with what changed, how it was verified, and any follow-up risks.
What is AI CLI Tools?

AI CLI Tools is a command generator for popular AI coding assistants. It helps you quickly build CLI commands with the right flags and arguments.

How It Works

Select an AI tool (Claude Code, Codex, or OpenCode), choose the options you need, and copy the generated command.

Common Use Cases
  • Generate Claude Code commands with specific models and effort levels
  • Build Codex CLI commands with sandbox and approval settings
  • Create OpenCode commands for automated tasks
Example Commands
Input: Claude Code
Output: claude -p "Explain this function"
Input: Claude Code
Output: claude -c -p "Continue with refactor"
Input: Claude Code
Output: claude --model sonnet --effort high
Input: Codex
Output: codex exec "Review my PR"
Input: Codex
Output: codex --sandbox workspace-write --search
Input: OpenCode
Output: opencode run "Review my PR"
Input: OpenCode
Output: opencode run --model anthropic/sonnet --continue
Input: OpenCode
Output: opencode serve --port 4096
Frequently Asked Questions

Is this an official tool?

No, AI CLI Tools is a community project. Please refer to official documentation for each tool.