Developer Builds Open Source Tool to Stop AI Models Answering Basic Terminal Commands

·3 views

A developer created an open source tool called Token Ninja after growing frustrated with wasting AI tokens on simple queries that a local terminal could resolve instantly. The tool intercepts prompts before they reach Claude, routing basic commands like git status or docker ps to the terminal instead of the language model. Built over a weekend, the project leverages Claude Code's UserPromptSubmit hook to enable zero-token interception at the prompt level. The developer found that the trickiest challenge was defining the boundary between requests suited for local execution and those genuinely requiring AI reasoning. Token Ninja is publicly available on GitHub under the profile oanhduong.

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)

Developer builds offline Android planner that reschedules your day without guilt

A developer has released Tiny Day, a free Android daily planner built with React Native and Expo that requires no account, no cloud sync, and collects no analytics. Users type their tasks in plain language each morning, and the app automatically parses them into a prioritized timeline while deliberately leaving free time and protecting meals. When plans go off track, a built-in repair engine reshuffles optional and flexible tasks around fixed commitments, showing users a preview before applying any changes. The app uses only local notifications, includes quiet hours and a privacy mode, and never carries unfinished tasks forward automatically. Its stated design philosophy centers on reducing guilt rather than enforcing productivity streaks or highlighting missed tasks.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer builds citation-only sales knowledge base that admits when it lacks answers

A developer created SalesWiki, an open-source starter kit for sales and marketing teams that stores compiled knowledge as typed Markdown files compatible with Obsidian. The system keeps raw evidence separate from conclusions and only returns answers backed by named sources, explicitly flagging gaps rather than generating unsupported responses. Changes to shared knowledge require a proposal and review step, keeping earlier evidence and the rationale for any update visible in Git history. SalesWiki is not a CRM or hosted SaaS product, but a governed knowledge base designed to help teams trace decisions back to verifiable evidence. A synthetic demo using a fictional Atlas Robotics deal illustrates how the tool surfaces reasons behind a suggested next action without removing human judgment from the process.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

OpenVPN Releases v2.7.1 and Connect 3.9.0: Why Version Tracking Matters for IT Teams

The OpenVPN community project released version 2.7.1 on September 3, 2026, while OpenVPN Connect for Windows 3.9.0 had been released earlier on June 8, 2026, with the two belonging to entirely separate release lines. Confusing the two version numbers can lead to unreliable upgrade planning, as they refer to different software components. A structured inventory approach is recommended, requiring teams to document the exact component name, release line, platform, configuration, and observation method separately. The method also distinguishes between an approved installer file and the version actually running on an endpoint, since the two represent different facts. Connect for Windows 3.9.0 introduces a pre-login connection feature that requires administrator setup and specific system profiles, meaning existing deployments cannot assume automatic compatibility.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Context7 MCP Tool Solves AI Code Assistants' Outdated Library Knowledge Problem

A developer using Astro 7 lost hours debugging code because their AI assistant kept suggesting APIs removed in Astro 6, unaware of the version in use. The root cause was the model's training data being frozen at a past cutoff, meaning it confidently produced correct answers for older Astro versions. Context7, a Model Context Protocol server, addresses this by fetching current library documentation and injecting it into the model's context before it generates a response. It requires no local installation and can be added to Claude or other editors via a single remote HTTP endpoint. After enabling Context7, the developer received accurate, up-to-date Astro API code on the first attempt, eliminating the version-mismatch failure entirely.

0 comments Read more at DEV Community