Tessl Agent automates recurring AI coding mistakes by scanning PRs and session logs

·1 views

Tessl has launched Tessl Agent, a tool designed to address the persistent problem of AI coding agents repeating the same errors without anyone having time to fix them. The tool continuously scans a repository's pull requests, coding agent session logs, and issue tickets to identify recurring error patterns. When a pattern is detected, it automatically creates a corrective skill and opens a pull request, while repetitive manual tasks are converted into GitHub Actions workflows. Tessl Agent is provider-agnostic and works alongside existing tools such as CodeRabbit and GitHub Actions rather than replacing them. Its goal is to reduce the need for manual intervention by scheduling automated improvement loops that run on user-defined triggers.

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 finds short break from stalled side project restores focus and progress

A college student building Aniotako, a personal anime tracking app with watchlists and episode notifications, paused development during exam season. After exams ended, persistent bugs and mounting frustration led them to set the project aside for over 10 days. Returning with fresh eyes, the developer found it easier to approach problems and resumed making meaningful progress. The experience prompted them to ask fellow developers whether pushing through or stepping back is the better strategy when a project becomes consistently frustrating.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

AI Is Replacing Traditional QA Roles, But Adaptive Testers Remain Valuable

The software testing profession is undergoing a major shift as AI tools like GitHub Copilot, Playwright, and Applitools increasingly automate tasks once handled by manual QA engineers, including test case generation, regression testing, and visual bug detection. The developer-to-QA ratio has grown from 3:1 in 2015 to roughly 10:1 in 2025, and Stack Overflow's 2025 Developer Survey found QA roles represent just 8.4% of respondents with a median salary of $57,442, well below other engineering roles. A May 2025 report by Sabrina Ramonov highlighted a technique called Automated Input Diversification, where large language models proactively generate test variants to catch bugs that standard unit tests miss. However, experts argue that QA professionals who evolve into roles such as Quality Architect, Risk Analyst, or AI Test Orchestrator still offer value AI cannot replicate, particularly in exploratory testing, business impact assessment, and cross-team communication. The core advantage of human QA lies not in running tests but in asking critical questions about edge cases and user behavior that AI tools are not yet equipped to anticipate.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Build a Safe Generic Table API Using Dapper and C# in 80 Lines

A developer tutorial on DEV Community demonstrates how to build a minimal, read-only generic API over a PostgreSQL database using Dapper and C# in roughly 80 lines of code. The approach avoids Entity Framework, code generation, and per-table endpoints by using a single generic endpoint with a whitelist of permitted tables and columns. Pagination is bounded and database access is restricted to read-only permissions, preventing accidental exposure of the entire database. Columns are returned in camelCase and JSON column types are automatically deserialized, reducing frontend parsing overhead. The template is presented as a practical internal tool rather than a fully dynamic and potentially insecure database browser.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Claude Skill 'tiny' Compresses Images via TinyPNG With One Command

A developer has built a Claude skill called 'tiny' that compresses PNG, JPEG, and WebP images using the TinyPNG API with a single instruction, replacing the original file with a smaller version. The tool demonstrated a real-world reduction of a cover image from 70.3 KB to 19.9 KB, a saving of over 71 percent. It supports single files, multiple files, and glob patterns, and safely leaves the original untouched if any error occurs during compression. A key feature is built-in usage tracking: TinyPNG's free tier allows 500 compressions per month, and the skill warns users before they exceed that limit to prevent accidental charges. The skill is installed by cloning a GitHub Gist into Claude's skills folder and registering a free TinyPNG API key.

0 comments Read more at DEV Community