SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Developer builds tool that auto-generates architecture explainers from AI-written code

0
·16 views

A developer has released 'show-me-the-build', an open-source skill compatible with Claude Code, Cursor, Codex, and other LLM-based agents. The tool analyzes any codebase or cloud deployment and generates a self-contained, interactive HTML presentation explaining what was built, how it works, and why each component exists. Architecture diagrams are rendered step by step, reflecting the actual order in which resources were created, with clickable elements that surface real reasons drawn from the code or infrastructure docs. The tool requires no server or build step and is aimed primarily at hackathon participants who need to quickly explain AI-assisted projects to judges. It can be installed via a single npx command and invoked directly inside a supported coding agent.

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 ·

When overflow: clip with overflow-clip-margin fits decorative effects, and when it doesn't

The article explains how to keep a container non-scrollable while letting decorative effects like shadows or focus rings paint past its edge. Using overflow: clip with a small overflow-clip-margin expands the paint clipping area without creating a scroll container. The author advises against this pattern for interactive or accessibility-critical content, where a true scroll container or layout change is needed. It also notes that shadow rendering varies between browsers, so visual checks in target environments are necessary. A minimal test fixture is recommended to validate the behavior before applying it to a real component.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Mergetober: Zendesk connector lets cognee agents answer from tickets and articles

A developer built a Zendesk connector for cognee so support agents can answer questions using the team's past tickets and Help Center articles. The connector supports email plus API token or OAuth access, the latter because Zendesk is retiring API tokens in 2026 and 2027. In a trial, deleting a ticket in Zendesk removed its content from the agent's answers after the next sync. Tickets sync incrementally with deletions handled as tombstones, while articles are checked against Zendesk's reported count to avoid forgetting content by mistake. Internal notes and email addresses are excluded by default to keep agents from quoting them as customer-facing information.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Idle Dev and QA Cloud VMs Waste Most of Their Billed Hours, Tool Reports

A developer post argues that dev and QA virtual machines are typically used about 60 hours a week but billed for all 168 hours. The author promotes Control Cloud, which reads cloud inventory via read-only provider API access to estimate overnight waste. On a demo organization, 16 running VMs at 12 hours nightly and $1.51 per hour amounted to roughly $6,627 a year in potentially recoverable spend. The platform also flags orphaned resources still billing, disks of stopped VMs, and savings from off-hours shutdown schedules. The post invites readers to describe how they currently manage overnight dev and QA resources.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

RIP.git Builds a Graveyard for Abandoned GitHub Repositories, Complete with AI Epitaphs

RIP.git is an open-source tool that turns a GitHub user's unmaintained projects into a static graveyard page. It pulls public, non-fork repositories through the GitHub API and classifies each as alive, dying, or dead based on days since its last push. Each repository gets a tombstone listing its status, time since death, last commit message, and an epitaph generated by the open-weight Gemma model via Ollama. Python handles all factual data, while Gemma is limited to writing the wording, with deterministic fallback text used when Ollama is unavailable. The project was built for Hacktoberfest's Open-Source AI Challenge and includes a fictional demo developer to avoid roasting real users.

0 comments Read more at DEV Community