Spend leftover AI quota on reviewable PRs
RemainderBot uses a founder's goals and task lists to write PRs with spare Claude and Codex quota, producing 111 PRs including call prep and prototypes.
- What they did
- The founder points RemainderBot at a goals file and task-list sources. It runs by hand or as a cronjob. In cron mode it checks remaining subscription quota and time until the weekly reset, picks a task, works on a new branch, and opens a PR. The founder reviews PRs on GitHub, and comments and merge decisions shape later tasks.
- How it went
- It produced 111 PRs, including customer call prep, a census of 31 GitHub orgs' agent setup files, and standalone prototypes for demos. The source gives no accuracy figures; the author asks users to report mistakes.
- Worth knowing
- Needs git, the claude or codex CLI, and an authenticated gh. A single small run with a 20-minute deadline costs a few percent of weekly quota.
Try it yourself with Claude
Read my goals in [path to goals file] and my task list in [path or link to task list]. Pick up to [number] small, well-defined tasks I could use help with, such as call prep notes, prototypes, or small fixes, and do each on its own branch and open it as a pull request in [repo] with a short description of what you did and why. Do not merge anything, and stop once you have opened the PRs and given me a list of them.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Oct 13, 2026
Related use cases
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Handle family errands and recipe sorting with Codex
A communications professional had Codex email her son's doctor for a refill and sort 240 recipes down to 62 family-fit ones into an index, spreadsheet and Drive folder.
Let Claude buy a game autonomously
The poster built payment infrastructure so Claude could complete a game purchase on its own when told to, in about 60 seconds.
Review Linux kernel patches
Google engineers launched Sashiko, an agentic AI code reviewer for the Linux kernel.
Custom agent
Triage failed GitHub Actions runs and self-heal flaky tests
A QA engineer's OpenClaw agent reads failed CI logs, classifies failures, posts analysis to GitHub and Slack and opens issues; flakiness fell from 8% to 0.8%, though it once wrongly altered an assertion.
Split a React Native to Swift migration across agents
Cognition's Jared Zoneraich uses Devin fan-out so child agents work in parallel on a migration split into six pieces.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.