Issue Agent

An AI Agent that quickly solves simple GitHub issues

---

Powered by Large Language Models (LLMs).

When a developer creates an issue in a repository and passes to this agent, it autonomously works to solve the issue and submits its results as a Pull Request on GitHub.

Design

• Data transmission is limited to the LLM provider and GitHub only
• Only execution of predefined tools is allowed, not arbitrary shell execution
• Creates a new working branch from the base branch for each task and submits pull requests - cannot push to or merge into the base branch

Quick Installation & Usage

Installation

• Your Machine
• GitHub Action

Usage

CLI

issue-agent help

$ issue-agent create-pr clover0/example-repository/issues/123 \
--base_branch main \
--model claude-sonnet-4-20250514

GitHub Action

1. Set up the workflow file in your repository.

name: Run Agenton:
issues:
# This workflow is triggered when an issue is labeled with 'run-agent'types:
- labeledcreate-pr:
if: ${{ github.event.label.name == 'run-agent' }}name: Create PRruns-on: ubuntu-lateststeps:
- name: Install Issue Agentuses: clover0/setup-issue-agent@v1
- name: configure aws credentialsuses: aws-actions/configure-aws-credentials@v4with:
role-to-assume: "arn:aws:iam::<AWS-ACCOUNT>:role/<ROLE-NAME>"role-session-name: run-agent-${{ github.run_id }}aws-region: "<AWS-REGION>"
- uses: actions/create-github-app-token@v2id: app-tokenwith:
app-id: ${{ secrets.TOKEN_APP_ID }}private-key: ${{ secrets.TOKEN_APP_PRIVATE_KEY }}
- name: Run Issue Agent Actionrun: | issue-agent create-pr ${GITHUB_REPOSITORY}/issues/${{ github.event.issue.number }} \ --base_branch main \ --model us.anthropic.claude-sonnet-4-20250514-v1:0 \ --aws_region us-east-1env:
GITHUB_TOKEN: ${{ steps.app-token.outputs.token }}

See setup-issue-agent for more details.

2. Label the issue with run-agent to trigger the workflow.

3. Issue Agent will create a Pull Request.

Documentation

Refer to the documentation for more details.

Key Features

• Fully Autonomous

  • Handles simple coding and documentation tasks without human intervention.

• Minimal Configuration

  • Easy to set up: runs locally or via GitHub Actions in minutes.

• Security

  • Limited scope: cannot execute arbitrary or unsafe code (only predefined and controlled functions).
  • The agent running in a container is only provided with the necessary environment variables, which minimizes the risk of credential leakage

Suitable Use Cases

Issue Agent makes life easier, especially for routine or repetitive tasks:

• Routine Development Tasks(e.g., basic migration, formatting, simple refactoring tasks)
• Documentation Maintenance(generate or update consistently formatted docs)
• Code Cleanup(e.g., batch updates, typo corrections, simple bug fixes)

Supported AI Models

We recommend Anthropic's Claude models for optimal performance:

Provider	Supported Models
OpenAI	gpt-4o, gpt-4o-mini
Anthropic⭐️	claude-sonnet-4-20250514, claude-opus-4-20250514, claude-3-5-sonnet-latest, claude-3-7-sonnet-20250219
AWS Bedrock⭐️	See AWS Bedrock section

AWS Bedrock

The following models are supported.

• claude-sonnet-4 (ModelID = us.anthropic.claude-sonnet-4-20250514-v1:0, Cross-region inference)
• claude-opus-4 (ModelID = us.anthropic.claude-opus-4-20250514-v1:0, Cross-region inference)
• claude-3-7-sonnet (ModelID = anthropic.claude-3-7-sonnet-20250219-v1:0)
• claude-3-7-sonnet (ModelID = us.anthropic.claude-3-7-sonnet-20250219-v1:0, Cross-region inference)
• claude-3-5-sonnet v2 (ModelID = anthropic.claude-3-5-sonnet-20241022-v2:0)
• claude-3-5-sonnet v2 (ModelID = us.anthropic.claude-3-5-sonnet-20241022-v2:0, Cross-region inference)
• claude-3-5-sonnet v1 (ModelID = anthropic.claude-3-5-sonnet-20240620-v1:0)
• claude-3-5-sonnet v1 (ModelID = us.anthropic.claude-3-5-sonnet-20240620-v1:0, Cross-region inference)