SuperAPI AI Directory
Partner with us中文AI newsFavorites
All toolsAI codingissue-ai-agent

issue-ai-agent

Issue AI Agent is an AI-driven GitHub Issue auto-classification and handling tool that runs as a GitHub Action. When a new Issue is created in a repository, it automatically classifies the Issue into categories such as bug, feature, question, docs, duplicate, invalid, or security, and applies the corresponding labels and priority (critical, high, medium, low). It can also detect duplicates by searching existing Issues and link potential matches, and send context-relevant reply comments based on the classification (such as asking for reproduction steps for a bug, or acknowledging receipt for a feature). In addition, it can handle follow-up comments. The entire process completes in about 8 seconds, powered under the hood by an LLM of the user's choice. Installation is done by adding a workflow file to the repository and configuring an API key, supporting Anthropic or OpenAI-compatible APIs, and the model and behavior can also be customized via a configuration file.

Open website

About this tool

Issue AI Agent is an AI-driven GitHub Issue triage Action, used to automatically complete Issue classification, labeling, and replies.

Purpose

When someone opens an Issue in your repository, Issue AI Agent automatically performs the following processing flow:

  • Classifies the Issue into a category (bug, feature, question, docs, duplicate, invalid, security)
  • Applies matching labels to it and determines priority (critical, high, medium, low)
  • Detects duplicates by searching existing Issues and links potential matching Issues
  • Sends a contextual comment—for example, bugs are asked for reproduction steps, features receive confirmation, etc.
  • Handles follow-up comments and replies with relevant information to user comments

The entire flow above completes in about 8 seconds, powered by the LLM you choose.

Usage

Step 1: Add the workflow file

Create in the repository .github/workflows/issue-ai.yml.

Step 2: Add an API key

Go to Settings > Secrets and variables > Actions > New repository secret:

  • Name:ANTHROPIC_API_KEY
  • Value: your Anthropic API key (available at console.anthropic.com)

Step 3: Open an Issue

After completing the above steps, the bot will automatically classify, label, and reply to new Issues.

Supported APIs

The bot supports any Anthropic or OpenAI API. Use the three inputs—<provider>-api-key、llm-provider and llm-base-url—to match your configuration.

Custom API Endpoint

If your provider offers an Anthropic or OpenAI-compatible API, you can point llm-base-url to its address.

Custom Model

If using a custom model, you can create .github/issue-ai.yml in the repository to specify it. This file is also used to customize behavior. The bot works out of the box with sensible defaults and requires no configuration file.

Configuration Reference

Action Inputs and Outputs configuration items are provided. At least one API key must be set. If neither is set, the bot will run in development mode and use simulated responses.

Key Design Decisions

  • Stateless—no database is used; configuration is read from each repository's .github/issue-ai.yml - Error-recoverable—each flow step catches errors independently, so a classification failure will not block replies
  • Security-first—input sanitization (zero-width characters, control characters, length limits) + explicit marking of untrusted data in prompts

Use Cases

Suitable for repository maintainers who want to automatically manage GitHub Issues, reducing the workload of manual classification, labeling, deduplication, and initial replies. Users only need to add a workflow file and configure an API key to automatically trigger processing when a new Issue is created.

Related tools

Browse category