DEV Community: Shekhar

Meet Hermes Agent: The AI That Writes Code & Crushes Jira Tickets!

Shekhar — Mon, 08 Jun 2026 02:09:42 +0000

Are you tired of manually translating Jira user stories into automation scripts? What if an AI agent could read your Jira ticket, write the required integration tests, execute them locally in your environment, verify the results, and automatically comment back on the ticket with the final status?
In this post, we are breaking down Hermes Agent, an incredible open-source AI developer agent from Nous Research that does exactly that. Based on an awesome walkthrough by Shekhar's Automation Insights, let's dive into what makes this tool a game-changer for developers and QA engineers alike!
What is Hermes Agent? 🤔
Hermes Agent is an open-source, MIT-licensed autonomous AI coding assistant. Unlike traditional AI extensions that just sit in your IDE and suggest code snippets, Hermes can actively interact with your system terminal, manage files, spin up test frameworks, and sync directly with project management suites like Jira.
It supports multiple LLM providers right out of the box, including:
Anthropic (Claude)
OpenAI (GPT-4o)
Google (Gemini via AI Studio)
Key Capabilities & Architecture 🌟
Autonomous Workflows: It safely creates environments, installs dependencies, and fixes bugs on its own.
Tool Integration Ecosystem: Equipped to manage file systems, execute arbitrary terminal tasks, and leverage browser automation tools like Playwright.
Bi-directional Sync: It reads task requirements directly from issue trackers and writes execution status logs back to them.
Step-by-Step Walkthrough: Watching Hermes Crush a Ticket 🛠️
Here is how the automation workflow unfolds in the live demo:

Quick Installation & Setup Getting started with Hermes Agent is remarkably simple. Using a single PowerShell command, you can install the agent directly:

irm https://hermes-agent.nousresearch.com/install.ps1 | iex

During the interactive setup wizard, you can opt for the Nous Portal (for a fast, OAuth-based keyless entry) or manually supply your own API keys. In the video, it is seamlessly configured using a Google Gemini model.

Initializing the Project Environment To put Hermes to the test, a fresh automation test project is bootstrapped locally inside VS Code using Playwright:

npm init playwright@latest

Automatic Code Generation & Verification The user instructs Hermes Agent to check a specific active Jira ticket titled "Login Functionality Verification" on a Scrum dashboard. Reading Requirements: Hermes connects to Jira, parses the description containing user credentials, target URLs, and specific evaluation benchmarks. Writing Code: It autonomously generates a complete Playwright test file (⁠test_login.spec.js⁠) matching the exact parameters of the ticket. Local Testing: Instead of guessing if the code works, Hermes executes the test script in the local terminal environment (⁠npx playwright test⁠).
Updating Jira and Closing the Loop Once the tests successfully pass locally, Hermes logs back into the Jira platform. It posts a detailed execution report directly into the ticket comments, updates the team on the verification steps, and completes the loop without requiring any manual copy-pasting from the engineer. Why This Is a Game-Changer for DevOps and QA 🏎️ 1 Eliminates Context Switching: Engineers don't have to bounce between Jira, VS Code, and terminal windows to document completion. 2 True Local Validation: Because the agent executes the code inside your local workspace, you can confidently verify that the generated scripts function perfectly. 3 Accelerated Testing Cycles: Writing boilerplate E2E and integration tests is offloaded entirely to the AI agent, freeing up engineers to focus on complex architecture. Watch the Full Walkthrough 📺 If you want to see the live step-by-step terminal installation, model configuration, and see Hermes Agent write and execute tests in real-time, check out the full video here:

🔗 Watch on YouTube: https://youtu.be/xXyL7h2Nx8c?si=grg47McRXlRHF42u

Don't forget to like and subscribe to Shekhar's Automation Insights for more cutting-edge automation tutorials!
*** Have you experimented with autonomous developer agents like Hermes in your workflow yet? Let's discuss in the comments below! 👇

Automate Test & Framework Setup in Seconds with agents-cli-automation

Shekhar — Sat, 21 Feb 2026 23:36:25 +0000

My portfolio : (https://shekhar-pro.vercel.app/about)

Demo :

If you’ve ever started a new automation project, you know the drill:

👉 Install dependencies
👉 Configure test runners
👉 Wire up folders
👉 Repeat the same setup… again and again

It’s repetitive, time-consuming, and honestly — not where you want to spend your energy.

That’s exactly the problem agents-cli-automation aims to solve.

🧠 What is agents-cli-automation?

agents-cli-automation is an npm CLI tool designed to bootstrap automation setups quickly using a guided, interactive workflow.

Instead of manually wiring everything, you can generate a ready-to-use structure with a single command — letting you focus on writing tests and building features instead of boilerplate.

⚡ Why it’s useful

Here’s where it shines:

Zero-friction setup → No need to memorize install steps

Consistent project structure → Standardized scaffolding across teams

Faster onboarding → New contributors can start instantly

Framework flexibility → Works with different automation approaches

In short: less setup, more shipping.

🛠️ Getting started
npx agents-cli-automation init

From there, the CLI guides you through configuration choices and generates the project structure automatically.

💡 Real-world use cases

Developers are using tools like this to:

Spin up Playwright + BDD frameworks quickly

Standardize QA automation across repos

Prototype testing setups for POCs

Reduce environment drift in teams

If you’re working in DevOps, QA, or developer productivity — this kind of CLI can save hours every sprint.

🤔 When should you try it?

You’ll get the most value if you:

✔ Frequently create new automation repos
✔ Want a repeatable setup process
✔ Are onboarding teammates often
✔ Prefer convention over configuration

🧩 Final thoughts

The ecosystem is moving toward automation-first developer workflows, and CLI scaffolding tools are becoming a key productivity layer.

agents-cli-automation fits right into that trend — helping teams eliminate repetitive setup and start building immediately.

If you’ve tried it (or built something similar), I’d love to hear how you’re automating your project setup workflows 👇

What’s the one step you wish you never had to do again?

Playwright Debugging, Supercharged by GitHub Copilot

Shekhar — Fri, 13 Feb 2026 06:39:04 +0000

My Portfolio :(https://shekhar-pro.vercel.app/)

What I Built

I built TestPilot AI, an AI-augmented Playwright test runner that automatically diagnoses failing tests and suggests fixes using the GitHub Copilot SDK and CLI.

The project’s main goal is to reduce the time developers spend debugging failing tests. Instead of manually inspecting error messages or stack traces, TestPilot AI sends failure details to Copilot, which analyzes the failure, identifies the root cause, classifies the issue (e.g., flaky test, locator change, API failure), and even provides a suggested code fix or improved test snippet.

For me, this project demonstrates how Copilot can act as a real development partner — not just generating code, but actively helping debug and optimize tests in real time.

Demo

Project Repository: (https://github.com/jancsp3/GHCPCLI-Shekhar)

Video Walkthrough:(https://youtu.be/JB_DcMVHVh8)

Run the AI test runner:

npm run test

A test fails intentionally.

Copilot SDK automatically analyzes the failure and prints a detailed diagnostic report in the terminal, including:

Root cause

Suggested code fix

Updated test snippet

Diagnostic report is also saved in /ai-diagnostics for review.

Example screenshot / snippet:

AI analyzing failure in: homepage should have correct title

--- COPILOT DIAGNOSTIC REPORT ---
Root Cause: The test expected 'Wrong Title' but the actual page title is 'Example Domain'.
Suggested Fix: Update the test to expect the correct title.
Improved Test Snippet:
await expect(page).toHaveTitle('Example Domain');

My Experience with GitHub Copilot CLI

I used GitHub Copilot CLI throughout the project to accelerate development and ensure high-quality outputs:

Project scaffolding:

gh copilot suggest "Create a Playwright TypeScript project structure with tests folder and custom fixtures"

Copilot generated a ready-to-use project structure, saving me time on setup.

AI diagnostic fixture creation:

gh copilot suggest "Create a Playwright fixture that runs after test failure and sends error details to Copilot SDK for analysis"

I refined the generated code to handle error reporting and session cleanup.

Prompt optimization:

gh copilot suggest "Improve this AI prompt for diagnosing Playwright test failures"

This helped produce more accurate diagnostic outputs.

Debugging and refactoring:

gh copilot explain ""
gh copilot suggest "Improve error handling in Playwright fixture"

Copilot provided explanations and suggested refactors, making iteration much faster.

Overall, Copilot CLI acted like a collaborative coding assistant — not only generating code but guiding me to improve prompts, debug errors, and optimize test workflows. It significantly reduced development time and made the testing process more reliable and intelligent.