Pullfrog is an open-source AI-powered GitHub bot in beta that runs entirely within GitHub Actions. It acts as an orchestration layer for asynchronous development by listening for webhooks and triggering AI agent runs based on configurable events such as new pull requests, issues, CI failures, and review submissions. It uses a bring-your-own-key approach, allowing connection to multiple LLM providers and switching models through configuration. API keys are stored in GitHub secrets, and agent runs execute in the repository’s GitHub Actions environment via a pullfrog.yml workflow. Setup requires installing the Pullfrog GitHub App and adding the workflow file. Agent capabilities include a purpose-built MCP server for git and GitHub operations, isolated shell command execution without sensitive environment access, and a headless browser tool for end-to-end testing and UI iteration.
"Pullfrog, an open-source AI-powered GitHub bot created by Colin McDonnell, is in beta, positioning itself as a model-agnostic alternative to CodeRabbit that runs entirely within GitHub Actions. McDonnell announced Pullfrog on May 12, 2026. The tool is designed as an orchestration layer for asynchronous development inside GitHub, listening for webhooks and triggering AI agent runs based on configurable events such as new pull requests, issues, CI failures, and review submissions."
"Unlike CodeRabbit, which operates as a hosted SaaS platform with its own AI models, Pullfrog takes a bring-your-own-key (BYOK) approach. Developers can connect any LLM provider including Anthropic, OpenAI, Google, Mistral, DeepSeek, and OpenRouter, switching between models with a single configuration change. All API keys are stored using GitHub's secret management system, and agent runs execute in the repository's own GitHub Actions environment via a dedicated pullfrog.yml workflow file."
"Getting started requires installing the Pullfrog GitHub App and adding the workflow file to a repository. Developers can then tag @pullfrog in any issue, pull request, or comment to trigger an agent run, or configure automated triggers from the Pullfrog console. The full setup guide is available in the official documentation."
"The agent ships with a purpose-built MCP server for performing git and GitHub operations such as creating pull requests, leaving reviews, reading CI logs, and managing issues. Shell commands run in an isolated subprocess without access to sensitive environment variables. A headless browser tool is also included out of the box, enabling the agent to run end-to-end tests, take screenshots, and iterate on UI without additional configuration."
Read at InfoQ
Unable to calculate read time
Collection
[
|
...
]