The Deska blog
Blog
Product notes, release highlights and how we build Deska: the one infinite canvas workspace for your editor, terminal, browser, notes and AI agents.
· 10 min read
The Second Opinion Pattern: One Agent Reviews Another
Learn how the second opinion pattern improves LLM reliability by using one agent to review another agent's code in a side-by-side developer workspace.
· 10 min read
SDK Ergonomics: An Agent Review Before v1
Learn critical SDK ergonomics principles for AI agents to ensure your library is developer friendly before reaching v1 stability.
· 10 min read
Agent Scope Creep: When the Fix Touches Forty Files
Learn how to detect and prevent agent scope creep in your development workflow by using better context management and hybrid agent strategies.
· 9 min read
JSON-LD for a Blog: The Structured Data Pass
Learn how to implement JSON-LD for a Blog to improve SEO and discover how developer tools like Deska streamline technical content workflows.
· 9 min read
Safe Defaults for Running Agents: A Defensive Checklist
Learn how to implement safe defaults for running agents in your development workflow. A technical checklist for sandbox security and autonomous execution.
· 10 min read
The Test Loop: Making Agents Verify Their Own Work
Learn how to implement a robust test loop to ensure AI coding agents produce reliable code and avoid hallucinations through automated verification.
· 12 min read
Closing the Loop: Agents That Run the Test Suite
Learn how agents that run the test suite improve software reliability by providing immediate feedback loops and autonomous bug fixing in AI workflows.
· 10 min read
Runbooks by Agent: Capture the 3am Knowledge
Learn how to use AI agents to automate runbook writing and maintenance during production incidents to preserve critical operational knowledge.
· 11 min read
Telemetry for Agent Runs: Designing Observability for AI Coding
Learn how to design telemetry for agent runs to debug complex AI interactions, monitor token usage, and ensure reliable autonomous coding workflows.
· 11 min read
Storing Agent Run Artifacts
Learn technical strategies for storing agent run artifacts, managing ephemeral files, and using local-first workspaces for better agent operations.
· 10 min read
The Agent Rewrote Code That Already Worked
Learn why AI agents rewrite working code and how to prevent it using local-first tools like Deska with multi-agent visibility.
· 9 min read
Asking an Agent to Review Your Own Code
Learn how to effectively improve code quality by asking an agent to review your own code within a local-first development environment.
· 11 min read
One Agent Reviewing Another's Work: Strategies for Automated Code Quality
Explore how one agent reviewing another improves code reliability through multi-agent workflows, self-correction, and specialized validation panels.
· 11 min read
Retry Strategies When the Model Returns Garbage
Learn effective retry strategies when the model returns garbage, focusing on structured output, validation loops, and multi-agent debugging for LLM reliability.
· 11 min read
Session Resume Across Agents
Explore how session resume across agents affects developer productivity and how local-first environments like Deska manage persistent context for AI workflows.
· 10 min read
Turning a Vague Bug Report Into a Repro
Learn how to go from a vague bug report into a repro using automated workflows and AI agents within your development environment.
· 11 min read
From Bug Report to Repro Script With an Agent
Learn how to automate bug reproduction using AI agents. Convert bug reports into reliable repro scripts with Deska, Claude Code, and local-first workflows.
· 10 min read
Release Notes From Commits, Minus the Noise
Learn how to generate release notes from commits for your next deployment. Automate clean, readable logs using conventional commits and developer tools.
· 10 min read
Regex You Inherited, Explained by Agent
Learn a comprehension recipe for regex you inherited explained by agent tools, using Deska to visualize and refactor complex legacy patterns safely.
· 11 min read
Refactoring a 3000-Line File With an AI Agent
Learn how to refactor a 3000-line file with an AI agent. Master decomposition strategies and multi-agent workflows for massive legacy codebases.