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.
· 9 min read
The CSV Import Feature Every App Eventually Needs
A deep dive into building the CSV import feature every app eventually needs, focusing on robust parsing, error handling, and developer workflows.
· 11 min read
Migrating CSS to Tailwind With an Agent
Learn a technical recipe for migrating CSS to Tailwind with an agent, leveraging AI coding tools and infinite canvas workspaces for a better workflow.
· 12 min read
Auditing Cron Expressions: What Actually Runs When
Master the art of auditing cron expressions to prevent silent failures. Learn how to validate schedules and verify executions in your dev environment.
· 10 min read
What a Feature Costs in Agent Tokens: Real Numbers Thinking
Calculate the agent cost per feature by analyzing token consumption, context windows, and effective unit economics for AI-driven development.
· 12 min read
Metering Agent Costs Per Session
Learn technical strategies for metering agent costs per session to control LLM expenses and optimize agentic workflows in production.
· 10 min read
Attributing Agent Costs to Projects and People
Learn strategies for attributing agent costs to projects and people to optimize your FinOps for AI development workflows and manage token spend effectively.
· 11 min read
CORS Errors, Explained and Fixed by Agent
Learn how to resolve common CORS errors using AI coding agents and local tools. A technical guide to preflight requests and cross-origin resource sharing.
· 11 min read
When Agents Copy Your Worst Existing Pattern
Learn why AI coding agents copy your worst existing pattern and how to fix technical debt using context-aware developer tools like Deska.
· 10 min read
From curl Command to Typed Client
Learn how to transform any curl command to typed client code using AI agents and local developer tools for faster API integration.
· 10 min read
CONTRIBUTING.md That Contributors Actually Follow
Learn how to write a CONTRIBUTING.md that contributors actually follow to scale your open source project and reduce maintenance burden.
· 11 min read
Long Sessions Go Bad: Recognizing Context Rot
Learn how to identify and prevent context rot in long sessions with AI coding agents to maintain high performance and code quality.
· 10 min read
Context Management for Long Agent Sessions
Learn effective context management for long agent sessions, including token compaction and infinite canvas strategies for AI development.
· 10 min read
Moving Your Config Between Agents
Learn strategies for moving your config between agents like Claude Code and OpenCode to ensure portability and consistency in your development workflow.
· 9 min read
Config File Schemas: Validation From Day One
Learn how to implement config file schemas to ensure validation from day one. Improve developer experience with JSON Schema and local-first tools like Deska.
· 12 min read
Watching Competitor Changelogs With a Weekly Agent Digest
Learn how to automate watching competitor changelogs using AI agents to generate structured weekly intelligence digests directly in your workspace.
· 11 min read
Should Agents Write Your Commit Messages?
Explore the debate on if agents should write your commit messages and learn how to implement an automated hygiene workflow for better git histories.
· 11 min read
Comment Spam: Teaching Agents Restraint
Learn how to stop comment spam and keep your codebase clean when working with AI coding agents like Claude Code and Codex.
· 10 min read
Comment Rot: Deleting Lies From Your Codebase
Learn how to identify and automate the cleanup of comment rot to improve code maintainability using AI agents and manual hygiene practices.
· 10 min read
Making Agents Match Your Coding Style
Learn how to guide AI agents to match your coding style using system prompts, linting rules, and contextual workspaces to ensure consistent codebases.
· 11 min read
Reading COBOL With an Agent: Comprehension Before Courage
Learn strategies for reading COBOL with an agent to modernize legacy systems safely using local-first tools and infinite canvas workspaces.