Skip to content

Small fix pull requests

Runs on a schedule and opens one small pull request per fix, each proven by a test that failed before and passes after. Changes collect on a dev branch and you decide when they ship.

EngineeringWorks with Claude Code · Cursor+1Community example
Build your promptSee requirements

What it does

  • Reads new errors, failing tests and support themes
  • Picks one small bug or gap and writes a failing test
  • Fixes it and opens its own pull request to a dev branch
  • Logs each run and files a question when it needs a decision

Your approval boundaries

Creates drafts

These actions always wait for your yes:

  • Merging into the main branch or deploying
  • Anything touching production data
  • Changes to billing, login or email code

You decide what happens next.

More about this workflow

This setup is for developers and small engineering teams who want fewer chores. It takes a repeat chore off your plate and hands back something you can check quickly.

First, it reads new errors, failing tests and support themes. Next, it picks one small bug or gap and writes a failing test. Then it fixes it and opens its own pull request to a dev branch. Finally, it logs each run and files a question when it needs a decision. It works with your git host, CI, monitoring and chat, each on its own login with limited access.

You get the prep work done and waiting for you. The agent stops and asks for your yes before merging into the main branch or deploying, anything touching production data and changes to billing, login or email code, and says plainly if a step cannot be undone.

“bug fixes with a failing-then-passing test as proof”
Matt Mazur @mhmazur · Sep 21, 2026
Before you startGit host · CI accessScoped permissionsClaude Code or Cursor
Explore required tools and model recommendations

Tools and access

  • GitHub

    Handles the git host and ci part of this job. Grant a fine-grained token, read code and write comments.

    API required: Yes. Needs a GitHub App or a fine-grained personal access token scoped to the repos it needs.

    Sign up: Create a GitHub account

  • Sentry

    Handles the monitoring part of this job. Grant read-only token.

    API required: Yes. Needs an auth token from your Sentry account.

    Sign up: Sentry

  • Slack

    Handles the chat part of this job. Grant a bot token in the channels it needs.

    API required: Yes. Needs a Slack app created in your workspace. It gets a bot token with the scopes you approve.

    Sign up: Create a Slack app

Official links only. Not sponsored, no affiliate links.

Model recommendations

Best pick

Claude Opus 5.5

Its maker points to it for long running coding work.

$4 per million input tokens, $20 per million output tokens.

Anthropic docs as of October 11, 2026

Cheaper backup

Claude Sonnet 5.5

Half the price per token and still strong for reviews and docs.

$2 per million input tokens, $10 per million output tokens.

Anthropic docs as of October 11, 2026

Our own picks. Not sponsored, no affiliate links.

Build your prompt

Add your details, then copy the prompt into your tool.

1Your details

Fill these in and the prompt updates as you type.

e.g. hourly

e.g. billing, auth

files or lines

2Your prompt

4 of 4 still empty
Fill these in before you run it: [YOUR REPO AND DEV BRANCH] [HOW OFTEN TO RUN: e.g. hourly] [AREAS IT MUST NOT TOUCH: e.g. billing, auth] [MAX SIZE OF ONE CHANGE: files or lines] You are the Small fix pull requests agent. Runs on a schedule and opens one small pull request per fix, each proven by a test that failed before and passes after. Changes collect on a dev branch and you decide when they ship. Use only the details filled in above. If any [BRACKETED] item is still blank or unclear, stop and ask for it first. Do not guess. What you do on your own: - Reads new errors, failing tests and support themes - Picks one small bug or gap and writes a failing test - Fixes it and opens its own pull request to a dev branch - Logs each run and files a question when it needs a decision What you must never do without a yes from a person: - Merging into the main branch or deploying - Anything touching production data - Changes to billing, login or email code Rules: - Before any step on the list above, stop and ask. Say in plain words what you want to do, who it affects, and whether it can be undone. - If something cannot be undone, say so first, in one short line. - Write drafts, not final actions. A person sends, pays, posts, or deletes. - Keep a short log of every action you took and every source you read. - If you are unsure, ask instead of guessing.
View run instructionsPaste into: Claude Code, Cursor, OpenAI Agents SDK

Your answers stay in this browser only.

Run your workflow

Connect your tools with only the access this setup needs, then follow these steps.

Steps

Do these in order. Test on old data before it touches live work.

  1. 1Connect GitHub with its own login, not yours. Grant a fine-grained token, read code and write comments.
  2. 2Connect Sentry with its own login, not yours. Grant read-only token.
  3. 3Connect Slack with its own login, not yours. Grant a bot token in the channels it needs.
  4. 4Paste the prompt where your harness takes instructions. See the note under the Copy button.
  5. 5Do a dry run on old or test data first. Check every draft and every stop point by hand before you let it near live work.
  6. 6Read its log after the first real run. Only widen access once a week of runs looks clean.

Safeguards and trust score

The score describes this example setup’s access, approval boundaries, ability to undo changes, and scope.

Access24 / 30

Writes drafts and notes inside your tools, nothing leaves on its own.

Human approval30 / 30

Waits for a yes on: merging into the main branch or deploying, anything touching production data, changes to billing, login or email code.

Undo and logs14 / 20

Proposes changes to live records, so the setup needs a change log and a way back.

Scope16 / 20

Runs on its own login across 4 tools. Fewer connections means a smaller blast radius.

How to raise it

  • Undo and logs: Stop it from editing live records. Have it write the change as a proposal you apply, and keep a log of every change with a way to roll it back +3
  • Scope: Cut it down from 4 connected tools to the 2 it needs most, each with its own limited login instead of your full account +2

Do all of these to reach 89, up from 84.

Want help getting there? Get an AI Trust Audit from Pixeldesigns

Reviewed by Trustcase (example setup). How scoring works

Embed badge

Show this trust score on your site or README. It links back here.

Trustcase trust score badge
Markdown
[![Trustcase trust score for Small fix pull requests](https://trustcase.dev/badge/small-fix-pull-requests.svg)](https://trustcase.dev/use-cases/small-fix-pull-requests)
HTML
<a href="https://trustcase.dev/use-cases/small-fix-pull-requests"><img src="https://trustcase.dev/badge/small-fix-pull-requests.svg" alt="Trustcase trust score for Small fix pull requests" height="20"></a>

More engineering workflows

  • New issue intake sorter

    Reads each new issue on your repo and sorts it as ready to work, needs more info, or too big. You get a clean queue and reporters get asked for missing details.

  • Merged changes digest

    Reads every pull request merged since the last run and writes a short digest of what changed. Each change is sorted by who it matters to and how big it is.

  • Morning QA report

    Runs a set of read-only checks on your codebase each morning and files one report issue. It starts as an observer and only gets more power when you grant it.

  • Upstream blocker recheck

    Once a month, re-tests the upstream bugs and missing releases your work is waiting on. Tells you which ones have cleared.