franklioxygen/agent-workflows — explained in plain English
Analysis updated 2026-05-18
Point an AI coding assistant at a specific workflow file, such as bug-fixing or code review, before starting a task.
Use the bundled workflow-automation skill so the assistant picks the right workflow file automatically.
Install skills for Codex covering security review, migration planning, or release preparation.
Compare workflow-guided runs against unguided runs using the included pass-rate study.
| franklioxygen/agent-workflows | bhartiyashesh/purelymailcalendar | biao994/docpaws | |
|---|---|---|---|
| Stars | 55 | 55 | 55 |
| Language | Python | Python | Python |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 2/5 | 4/5 | 3/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Agent Workflows is a library of structured step-by-step guides for AI coding tools and human engineers to follow when working on software projects. Instead of giving an AI assistant a single open-ended instruction and hoping it does the right thing, this library provides separate workflow documents for specific situations: starting a new project, adding a feature, fixing a bug, reviewing code, responding to a production incident, refactoring, and cleaning up technical debt. The core idea is that following a defined process with checkpoints produces better and more consistent results than ad hoc prompting. The README describes this as helping absorb "silent base-model quality drift" meaning that even if the underlying AI model becomes less reliable over time without an obvious announcement, the structured steps still catch mistakes before they slip through. The repository includes a small study using metrics like task pass rate, regression-free rate, and no-rework rate, comparing workflow-guided runs against unguided runs across six test tasks. Each workflow is a Markdown file that an AI coding assistant can read directly. Users either point the assistant to the relevant file manually, or use a bundled skill called workflow-automation that reads a task description and picks the right workflow file automatically. The library also includes shared files covering safety rules, preflight checks, and common conventions that all workflows reference, so each individual workflow document stays focused. Beyond the workflow documents, the repository ships several installable skills for tools like Codex: security review, test strategy design, migration planning, performance review, documentation maintenance, and release preparation. Each skill folder contains a metadata file and instructions, and is designed to be copied into a Codex skills directory and then invoked with a task description. The README notes the library is available in English and Simplified Chinese, and points new users to a separate getting-started guide included in the repository.
A library of Markdown workflow guides that AI coding assistants and engineers follow for tasks like starting a project, fixing a bug, or reviewing code.
Mainly Python. The stack also includes Markdown, Python.
No license is stated in the provided text.
Setup difficulty is rated easy, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.