In the age of AI-assisted coding, the bottleneck has shifted from implementation to specification. Developer Waldov, in a recent post on DEV.to, argues that "writing the ticket is the work," and everything after it is merely execution. For developers using agents like Claude Code, the quality of the output is strictly capped by the quality of the input spec. A vague brief yields vague software, whereas a rigorous ticket forces design decisions, edge case identification, and verification strategies to be resolved before a single line of code is generated.

Scope Drift Is the Enemy

The core discipline involves treating every task, bug, and feature as a distinct ticket before touching the editor. Waldov notes that most bugs originate from "while I'm in here, I should also fix X" mentality, which introduces scope creep. By mandating that adjacent issues trigger a new ticket, the current workflow remains clean and focused. This approach was tested on NfcPress, an NFC business card platform, where a bulk order bug revealed two distinct root causes: background job termination by Vercel and improper sorting logic. Writing the ticket forced a proper diagnosis, turning a vague symptom into two specific, documentable failures.

From Bugs to Regression Harnesses

The workflow creates a self-reinforcing loop where tickets evolve into tests. After closing the bug ticket (NFC-57), Waldov immediately opened a new ticket (NFC-58) to create a Playwright test covering the identified failure modes. This transforms transient bugs into permanent guards. Furthermore, periodic retrospectives analyze closed tickets to generate broader smoke tests and "goldens"β€”one-line statements of required behavior. After 38 tickets, a retro produced new tests for evals/smoke.ts and evals/product-evals.md, ensuring that historical knowledge is encoded into the deployment harness.

The Kanban Is Just a View

Crucially, the system relies on markdown files as the source of truth, not the visual kanban board. Each ticket is a .md file with frontmatter for status and a detailed spec for context, scope, and verification. A sync daemon generates the board from these files, meaning the visual interface is disposable while the knowledge base persists. This structure allows developers to grep, count, and reference tickets programmatically, turning the ticket history into a searchable context window for future AI interactions.

Key Takeaways

  • Spec First, Code Later: Design decisions and edge cases must be resolved in the ticket, not during implementation.
  • Bugs Are Tickets: Never fix a bug inline; diagnose it via a ticket to ensure root causes are documented and tested.
  • Markdown Is the Database: Use .md files as the source of truth for project state, making the data greppable and version-controlled.
  • Retros Build Harnesses: Periodically review closed tickets to generate automated smoke tests and manual golden checklists.
  • Context Compounds: Each new ticket inherits the test suite and goldens from previous work, creating a safety net that grows with the project.

The Bottom Line

Waldov’s method proves that in AI-assisted development, the human’s job is no longer writing code, but writing unambiguous, testable intent. If you aren’t treating your tickets as the primary engineering artifact, you’re letting the AI guess your design. Stop treating tickets as administrative overhead. In 2026, the ticket is the code; everything else is just compilation.