SShortSingh.
Back to feed

Developer builds a local Gemma-powered CLI that picks three priority tasks from a to-do list

0
·2 views

For Hacktoberfest's third week, a developer built sloth-triage, a roughly 70-line Bun command-line tool that asks a model to select the few tasks worth doing today from a text list. The tool sends the list to an open-weight model, with Gemma served locally through Ollama as the default, so the data stays on the user's machine and no account or API key is needed. Its prompt favors deadlines and tasks that worsen if ignored, and it returns a short ordered plan noting what can be skipped or deferred. The author notes limitations: small models can mistake urgency for loudness, and the tool does not yet read calendar data. The project is available on GitHub and was submitted to the Hacktoberfest Open-Source AI Challenge.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Jira Cloud Adds Native Formula Fields for Calculated Custom Values

Jira Cloud now includes a native formula field that lets users create calculated custom fields, available on all editions and in both company-managed and team-managed spaces. The feature responds to a long-standing request, JRACLOUD-61966, which was open for over eleven years before being marked as fixed. Formula fields cannot currently reference select, priority, or other formula fields, and empty formulas appear as zero in JQL searches. Jira Data Center does not yet offer this capability, as its equivalent request remained under review at the time of reporting. The guide walks users through creating a formula field, selecting its space scope, writing the formula, and verifying results.

0
ProgrammingDEV Community ·

Developer builds local AI tool to clear WhatsApp forwards from a parent's phone

A developer built Good Morning Sweep, an open-source tool that removes WhatsApp forwards, duplicates and accidental photos from a phone without uploading anything. The tool connects the phone to a laptop, where the open Gemma 4 model reviews each picture and describes its content. Simple rules then suggest deletions, while photos of people, documents and uncertain items are never suggested for removal. The user reviews every suggestion before anything is deleted, and the phone verifies each file against its laptop backup first. In a personal test, the developer deleted 1,641 files by hand in 35 minutes, and the tool was built mainly for a parent's phone.

0
ProgrammingDEV Community ·

Hydroniq: Open-Source Platform Helps Communities Report and Track Water Problems

Developers have released Hydroniq, an open-source water intelligence platform that lets communities submit and organize reports about leaking public taps, broken handpumps, and other water infrastructure issues. The platform tracks reports through Pending, In Progress, and Resolved statuses and offers a live dashboard and analytics on the most frequently reported problems. It uses Gemma 3 4B via Ollama to suggest issue categories, summarize reports, flag missing details, and propose next steps, though these outputs are advisory and require human review. The current live demo runs on a React frontend, FastAPI backend, and cloud MySQL database, while AI analysis works only in local development for now, with hosted inference planned as a future improvement.

0
ProgrammingDEV Community ·

Developer Builds Local AI Agent That Summarizes WhatsApp Groups Overnight

A developer created WhatsApp Digest, a locally run AI tool that reads exported WhatsApp group chats and produces a short daily summary in one markdown file. Users set it up once, and a scheduled job runs it automatically each morning to surface important announcements across groups. The system relies on Ollama to serve models such as Gemma 4 E4B, with Qwen3 1.7B offered for lower-spec machines, and requires no internet connection or cloud service after setup. The developer emphasizes privacy, arguing that group messages with personal information should not be sent to external APIs. The tool is free and runs on the user's own computer.