DEV Community: Atharv Vhanalkar

Dummy project

Atharv Vhanalkar — Fri, 23 Oct 2026 21:37:18 +0000

This is a submission for the MLH x DEV Writing Challenge

What I Built

Demo

Partner Technologies

Hackathon Experience

LazyStandup: Automating Daily Engineering Standups with Git Telemetry

Atharv Vhanalkar — Fri, 23 Oct 2026 21:36:31 +0000

🦥 LazyStandup: Automating Daily Engineering Standups with Git Telemetry

Every morning, engineering teams across the world run into the same familiar chore:

"What did you finish yesterday? What are you working on today? Any blockers?"

Most of us either frantically skim through our terminal bash history two minutes before the sync, or write vague bullet points like "Worked on tickets" or "Investigated bug".

In honor of International Sloth Day and Hacktoberfest's "Slow and Steady" theme, I built LazyStandup: a lightweight, privacy-first CLI tool that automatically turns your local git commits, diffs, and branch activity into crisp, structured daily standup notes.

💡 The Problem & The Lazy Payoff

As Semtione pointed out in StandupBot and Gitmore discussed regarding dev reporting, developers already track their progress automatically—it lives right inside their git history.

Manually retyping commit messages into Slack or team wikis is pure repetitive friction. LazyStandup eliminates that cognitive tax entirely:

Yesterday's Progress: Extracted directly from recent commits, stripped of git noise, and classified into [Features], [Bug Fixes], [Refactoring], or [Documentation].
Today's Plan: Inferred from active unstaged/staged diffs and branch modifications.
Blockers: Defaults to clean status or accepts custom inline flags.

🛠️ Architecture & Local-First Privacy

Many automated reporting bots require connecting third-party cloud services to private codebases. LazyStandup is designed with local sovereignty at its core:

┌─────────────────────────┐
│ Local Git Repository    │
│ (.git/logs, status diff)│
└───────────┬─────────────┘
            │
            ▼
┌─────────────────────────┐
│ Commit Parser & Filter  │ ──> Categorizes (feat, fix, chore, docs)
└───────────┬─────────────┘
            │
            ▼
┌─────────────────────────┐
│ Standup Engine          │ ──> Synthesizes Yesterday / Today / Blockers
└───────────┬─────────────┘
            │
            ▼
┌────────────────────────────────────────┐
│ Presentation Formatter                 │
│ (Markdown | Slack mrkdwn | JSON Schema)│
└────────────────────────────────────────┘

Standard Library Only: Zero heavy dependencies; runs on pure Python 3.8+.
Offline & Air-Gapped: Your internal commit messages and proprietary code paths never leave your local workstation.
Multi-Platform Output: Renders formatted Markdown for GitHub/Notion, Slack mrkdwn (with status emojis), or structured JSON.

🚀 Quick Demo & Usage

Run it inside any repository directory:

# Standard Markdown format
lazystandup

# Slack formatted output ready for pasting
lazystandup --format slack

# Filter by time range or author
lazystandup --since "yesterday" --format slack

Sample Slack Output

*Daily Standup Update*

*Yesterday:*
• [Features] Implemented git log extraction parser (`e9f1a23`)
• [Bug Fixes] Resolved timestamp zone offset in git range (`4c82b10`)

*Today:*
• WIP on `lazystandup/cli.py` (M)
• Wrap up active pull requests & review updates

*Blockers:*
• None :white_check_mark:

🌍 Why Open Innovation Matters

Automation tools should respect developer privacy. By relying on open, inspectable local tooling rather than opaque cloud aggregators, developers keep full ownership of their telemetry while still saving valuable minutes every morning.

Check out the project repository: GitHub - AtharvVhanalkar/lazystandup.

Hello DEV! 👋 Building AI Agents with Antigravity & DevRelay

Atharv Vhanalkar — Fri, 23 Oct 2026 21:15:47 +0000

Hello DEV Community! 👋

Hey everyone! I'm Atharv Vhanalkar, an AI/ML student at Ajeenkya DY Patil University in Pune, India.

I'm excited to officially introduce myself here on DEV! I've been diving into AI agent workflows, participating in hackathons, and exploring open-source tools that streamline building and debugging.

What I'm Exploring

AI Agent Workflows & Tooling: Connecting agentic assistants like Google Antigravity with MCP (Model Context Protocol) gateways.
Hackathons & Open Source: Gearing up for MLH hackathons and active Hacktoberfest challenges.
Building in Public: Sharing devlogs, postmortems, and lessons learned directly with the developer community.

Here is a glimpse of an interactive agent session saved via DevRelay:

Connecting Antigravity with DevRelay and MLH

You

Connect Antigravity with DevRelay and verify MLH account connection

Agent

Configuring DevRelay MCP server and verifying MLH connection.

▸ mlh_connection_status {}

Input

{
}

Output

Connected as Atharv Vhanalkar

Looking forward to connecting with fellow builders, learning together, and sharing more project breakdowns soon! 🚀

SnapTicket: Turn Client-Side Crashes Into Instant Engineering Tickets

Atharv Vhanalkar — Fri, 23 Oct 2026 20:00:22 +0000

🎫 SnapTicket Agent

Autonomous Incident Triage & Telemetry Fusion Engine

⚡ Executive Summary

SnapTicket bridges the critical chasm between chaotic client-side runtime crashes and structured engineering issue trackers. By fusing low-level macOS diagnostic telemetry (.ips reports) with visual error screenshots, SnapTicket leverages Gemma 4 to eliminate ambiguous bug reports, diagnose root causes in seconds, and dispatch production-ready tickets seamlessly.

🛑 01 // The Friction: The "It Crashed" Dilemma

When an application fails in production, teams face a massive information gap:

The User Side: End-user reports are notoriously vague ("the screen froze"), lacking essential stack traces, exception codes, and environment metadata.
The System Side: Native OS crash logs (.ips files) contain invaluable post-mortem data—memory registers, faulting thread dumps, and dyld images—but are buried in cryptic system directories (~/Library/Logs/DiagnosticReports/).
The Manual Bottleneck: Manually correlating when a user captured a visual glitch with which thread faulted in the system logs wastes hours of engineering triage time.

🧠 02 // The Solution: Autonomous Telemetry & Visual Fusion

SnapTicket automates this pipeline entirely through a local-first, multimodal approach:

┌────────────────────────┐      ┌──────────────────────────────────┐
│ Visual Glitch Capture  │  +   │ macOS Telemetry (.ips Log Trace) │
│ (Retina Screenshot)    │      │ (Faulting Thread, Exception Code)│
└───────────┬────────────┘      └────────────────┬─────────────────┘
            │                                    │
            └─────────────────┬──────────────────┘
                              ▼
        ┌──────────────────────────────────────────────┐
        │        Gemma 4 Multimodal Reasoning          │
        │  - Correlates UI error state with log dumps  │
        │  - Identifies target app category (Native/3rd)│
        │  - Formats reproduction steps & diagnostics  │
        │  - Dynamically deduces canonical portal URL  │
        └─────────────────────┬────────────────────────┘
                              ▼
        ┌──────────────────────────────────────────────┐
        │       Privacy & Consent Control Studio       │
        │   [Attach Visual Evidence] vs [Text Only]    │
        └─────────────────────┬────────────────────────┘
                              ▼
                 Automated Dispatch Pipeline
                 ├── Channel A: Authenticated GitHub REST API
                 └── Channel B: Human-Assisted Web Portal Co-Pilot

🛠️ 03 // Core Architecture & CapabilitiesPassive Telemetry SensingNon-intrusively scans ~/Library/Logs/DiagnosticReports/ for recent incident dumps, extracting exact exception codes and faulting process names instantly.Context-Aware ClassificationDeduces whether an app is a Native macOS App (routing toward Apple Feedback Assistant) or a 3rd-Party App (routing to respective repositories).Multimodal Inference (Gemma 4)Downscales Retina captures to optimize memory headroom, fusing visual pixel evidence with stack trace telemetry into a structured Markdown report.Binary Privacy Consent GateEnables total data custody with a binary toggle: attach visual proof or strip images completely for an air-gapped, text-only diagnostic dispatch.Dual-Channel DispatcherPushes clean Markdown tickets directly via authenticated GitHub REST APIs or stages them via a browser co-pilot paired with native pbcopy clipboard fallback.📊 04 // Technical SpecificationsComponentTechnology / ImplementationRoleCore AI EngineGemma 4 (via local Ollama or Google AI Studio)Multimodal visual and log cross-reasoningBackend FrameworkFastAPI + Uvicorn (Async Python)Non-blocking REST API endpointsOS Telemetry ParserCustom Python log sensor (core/log_sensor.py)Extracts .ips exception vectors & threadsFrontend UIVanilla HTML5 / CSS3 (Zero-dependency)Responsive dashboard with animated progress trackingDispatch PipelineGitHub REST API & Playwright / pbcopyAutomated issue creation & browser autofill🚀 05 // Quickstart GuideGet the project running locally in seconds:Bash# 1. Clone the repository
git clone https://github.com/AtharvVhanalkar/snapticket-agent.git
cd snapticket-agent

2. Install dependencies

pip3 install -r requirements.txt

3. Configure environment

cp .env.example .env

4. Launch the agent server

python3 main.py
Open http://127.0.0.1:8000 in your browser to start triaging.🔒 06 // Security & Data SovereigntyLocal Sovereignty: When running in local mode via Ollama, diagnostics, logs, and screenshots never leave your machine.Credential Protection: API keys and tokens are restricted strictly to local .env variables and excluded from version control.

nope

Atharv Vhanalkar — Wed, 21 Oct 2026 18:37:59 +0000

nice post

Atharv Vhanalkar — Wed, 21 Oct 2026 18:35:35 +0000

hi

Atharv Vhanalkar — Wed, 21 Oct 2026 12:53:07 +0000