Skip to main content

Dependency-free prompt scheduler for AI rate-limit resets.

Project description

PromptQueue

CI License: MIT Python 3.10+ PyPI

Schedule AI prompts for the moment your limits reset.

PromptQueue is a tiny, dependency-free prompt scheduler for Claude, Codex, ChatGPT, Gemini, Copilot, Cursor, Antigravity, and anything else you can launch from your machine.

When an AI tool says "try again at 7:30 PM", do not keep the tab open, set a reminder, or paste the same prompt later. Queue it now. PromptQueue waits, opens or focuses the target app, pastes the prompt, and submits it when the time arrives.

promptqueue add 19:30 claude finish the migration plan
promptqueue run

One Python file. Standard library only. No accounts, no server, no private APIs.

PromptQueue terminal demo

Why This Exists

AI rate limits waste the worst kind of time: attention.

You already know what you want to ask next. The only problem is the clock. PromptQueue turns "come back later" into a queued job you can trust.

Use it when:

  • Claude, Codex, ChatGPT, Gemini, or Copilot tells you to wait for a reset
  • you want a long prompt to run after you sleep
  • you want a coding agent to resume work when capacity returns
  • you want a dead-simple local queue instead of another SaaS dashboard

If this saves you one context switch, it has already done its job.

30 Second Demo

Queue a Claude prompt for 11:30 PM:

promptqueue add 23:30 claude write a first draft of the launch email
promptqueue run

Queue a Codex CLI prompt:

promptqueue add 23:30 codex-exec add tests for the queue runner
promptqueue run

Queue a multiline prompt safely:

@'
Review this repo.
Find the smallest useful next improvement.
Then implement it.
'@ | promptqueue add --stdin 23:30 claude
promptqueue run

View what is waiting:

promptqueue list --full
promptqueue show JOB_ID

Supported Targets

antigravity  Google Antigravity UI
chatgpt      ChatGPT web UI
claude       Claude web UI
claude-code  Claude CLI
codex        Codex desktop UI
codex-exec   Codex CLI
copilot      Copilot web UI
copy         clipboard only
cursor       Cursor UI
gemini       Gemini web UI

Run this to see the built-in aliases on your machine:

promptqueue targets

UI targets open or focus the app, click near the composer, paste, then press Enter by default.

CLI targets run the command directly with the prompt as one argument.

Agent Ready

This repo includes instruction files for common AI coding tools:

AGENTS.md                         Codex and general agents
CLAUDE.md                         Claude Code
GEMINI.md                         Gemini CLI
.cursor/rules/promptqueue.mdc     Cursor
.agents/skills/promptqueue/       Antigravity-style skill

That means you can tell an agent:

My limits reset at 7:30 pm. Schedule this prompt for Claude:
"Continue the refactor and run the tests."

The agent should queue it with PromptQueue instead of making you remember.

Requirements

  • Python 3.10+
  • Windows for GUI paste/submit automation
  • macOS/Linux work for queue management, URLs, clipboard fallback, and CLI targets, but GUI paste currently uses Windows APIs

Install

Install from PyPI:

pip install promptqueue
promptqueue selftest

Or clone the repo and run the single Python file:

git clone https://github.com/AtharvaMaik/PromptQueue.git
cd PromptQueue
python promptqueue.py selftest

After pip install, use promptqueue. When running from a clone without installing, use python promptqueue.py.

PromptQueue stores jobs in:

%USERPROFILE%\.promptqueue.json

Override the queue path when you want an isolated queue:

$env:PROMPTQUEUE_FILE="C:\path\queue.json"

How It Works

  1. add writes a job to the local queue file.
  2. run checks for due jobs.
  3. When a job is due, PromptQueue copies the prompt to the clipboard.
  4. For UI targets, it opens or focuses the app, clicks the composer, pastes, and optionally submits.
  5. For CLI targets, it launches the command directly.
  6. Attempts, failures, retries, and history stay visible in the queue file.

If the machine wakes up after the scheduled time, run catches overdue jobs.

Commands

add

Queue a prompt.

promptqueue add [options] WHEN TARGET PROMPT...

WHEN can be a local time like 23:30 or an ISO-ish datetime like 2026-06-24 01:15.

Options:

--click bottom                 Click near the bottom of the focused window before paste.
--click x,y                    Click x/y pixels from the target window top-left.
--command-template TEMPLATE    Run a custom command; use {prompt}.
--delay SECONDS                Wait before paste/submit. Default: 5.
--file FILE                    Read prompt text from a UTF-8 file.
--max-attempts N               Attempts before marking failed. Default: 5.
--no-submit                    Paste only; do not press Enter.
--pre-keys KEYS                Windows SendKeys before paste, e.g. "{ESC}".
--retry-base SECONDS           Initial retry delay. Default: 30.
--stdin                        Read prompt text from stdin.
--window TITLE                 Focus a window whose title contains TITLE.

Examples:

promptqueue add 23:30 claude summarize this paper
promptqueue add --window Codex 23:30 copy this lands in Codex
promptqueue add --no-submit 23:30 claude paste this but do not send it
promptqueue add --file prompt.txt 23:30 claude
"prompt from stdin" | promptqueue add --stdin 23:30 copy
promptqueue add --command-template "claude -p {prompt}" 23:30 command review this repo

run

Start the queue worker:

promptqueue run

Check once and exit:

promptqueue run --once

Poll faster or slower:

promptqueue run --poll 5

The runner must be alive when jobs are due.

list

Show queued jobs:

promptqueue list
promptqueue list --all
promptqueue list --full

show

Show one job as JSON:

promptqueue show JOB_ID

remove

Delete one job:

promptqueue remove JOB_ID

windows

List visible Windows window titles. Use this to find the right --window value.

promptqueue windows

targets

List built-in target aliases:

promptqueue targets

selftest

Run the built-in smoke test:

promptqueue selftest

Retries And History

Every job stores attempts and history. Failures stay visible in list and show.

promptqueue add --max-attempts 8 --retry-base 60 23:30 claude retry this more patiently
promptqueue list --all
promptqueue show JOB_ID

Backoff is exponential and capped at one hour.

Notes

PromptQueue intentionally does the boring thing: it uses local files, the clipboard, app launching, and keyboard paste. That keeps it portable and inspectable.

For GUI apps, it does not use private app APIs. If one app misses the composer, adjust --window, --click, --delay, or --pre-keys.

Contributing

Issues and PRs are welcome. Start with CONTRIBUTING.md and CODE_OF_CONDUCT.md.

The best contributions are small and practical: a better target alias, a failing selftest for a real bug, clearer setup docs, or a platform-specific paste improvement.

License

MIT. See LICENSE.

Star the repo if it saves you from waiting around for an AI limit reset. That is the whole point.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

promptqueue-0.1.0.tar.gz (13.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

promptqueue-0.1.0-py3-none-any.whl (12.6 kB view details)

Uploaded Python 3

File details

Details for the file promptqueue-0.1.0.tar.gz.

File metadata

  • Download URL: promptqueue-0.1.0.tar.gz
  • Upload date:
  • Size: 13.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.6

File hashes

Hashes for promptqueue-0.1.0.tar.gz
Algorithm Hash digest
SHA256 13649301550fb119a47dcce32f6b18ed7214cd7205834bf3255a6ff7c3008f7c
MD5 c8c754998684d89b7805ec787590d02f
BLAKE2b-256 7df806ae2f9d761fe9659891e5305e16a5a8f7f89f9252fbfdc587908f1c3607

See more details on using hashes here.

File details

Details for the file promptqueue-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: promptqueue-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 12.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.6

File hashes

Hashes for promptqueue-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 e906280ea34284addb5a4c5bc761b1c4941576679211d59c997fda4f1434de03
MD5 1f21692d3355d9cca5a54b266e1f08ed
BLAKE2b-256 d933e4ff0905650479e408496191395e159c8a1ea83c2385768979c8d68dfd6d

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page