Achieving Trusted Code in the Agentic SDLC
summary
Live demo copies an Express REST API prompt into Claude Code, analyzes the generated task-tracker with SonarQube and Sonar MCP (shows 65 issues to zero), then packages the prompt into a Tessl tile using the Tessl tile creator, adds rules and documentation tiles (libraries like Express and Zod), enforces patterns such as parameterized queries, rebuilds the app, and Sonar reanalysis reports zero issues, demonstrating a repeatable workflow that turns vulnerable AI output into production-ready code.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:17AI Coding Assistants Need Better Context
- 0:46Testing a Viral “Uber Prompt” for an Express API
- 1:57Building a Task Tracker App with Claude Code
- 2:54Checking the Generated App in SonarQube
- 3:36Sonar MCP Reveals Serious Security Issues
- 5:02Why Random AI Prompts Are Not Enough
- 5:26Turning the Prompt into a Tessl Skill
- 6:13Tessl Review Fails the Skill on Quality and Security
- 8:20Rebuilding the Prompt with Tessl Tile Creator
- 11:15Adding Rules, Docs, and Reusable Prompt Structure
- 13:46Rebuilding the App with the Reviewed Tile
- 15:28From 65 Issues to Zero, Then Publishing the Tile
related talks
36mAgentic CI/CD: from Pull Request to Production without the Paper Cuts
33mUnlocking agent beast mode with Tessl
01h 06mHow I Tamed Claude
01h 00mHow I Tamed Claude
46mThe amplified developer: zero to shipped in 30 mins with AI Coding Ass
47mThe Renaissance Developer: Building Better, Safer, & Quicker with AI-Powered Spec-Driven Development
33mShipping Fast, Breaking Silently: Why AI-Generated Code Needs a Second Opinion
36mCode Complete: The Day AI Writes Your Next App!