Grok Bot engineering lead calls for deleting the vast majority of code comments, saying they AI solidify temporary patches

Article

Grok Bot engineering lead calls for deleting the vast majority of code comments:AI turning temporary patches into inherited rules

Beating AI Breaking News, Grok Bot engineering lead Lauren Tan called for deleting the vast majority of comments in code repositories. She believes that Coding Agent likes to imitate existing code. A comment explaining a temporary fix may be treated by subsequent Agents as the correct approach and copied repeatedly, ultimately spreading outdated patches and workarounds throughout the project.

Tan previously revealed in an interview that Dune, the development framework used by Grok Bot, has already banned code comments. She estimated that, of the AI comments she has encountered generated by 99%,PR when engineers merely asked a API not to use a specific implementation, the Agent instead wrote the instruction into a comment, leaving it for subsequent Agents to treat as a permanent rule.

For this reason, Tan recommends using the `/no-comments` Skill in the open-source pstack toolkit. It invokes a dedicated Agent to review comments, identify outdated explanations and temporary workarounds, and then address the root problem through refactoring. For example, an architectural requirement originally written in a comment can be changed into an automatically enforced Lint rule, or the type system can directly prevent an incorrect implementation.

However, necessary explanations may still be retained for external constraints that cannot be resolved independently, such as browser compatibility and third-party dependencies. Public https://m.theblockbeats.info/flash/371767

Key points

Grok Bot engineering lead Lauren Tan called for deleting the vast majority of comments in code repositories.
Tan believes that Coding Agent may cause comments explaining temporary fixes to be treated as correct practices and copied repeatedly.
Tan said that Dune, the development framework used by Grok Bot, has already banned code comments.
Tan estimated that, of the comments she has encountered AI generated, 99%were explaining trivial historical background.
Tan recommends using the `/no-comments` Skill in the open-source pstack toolkit to review comments.
The report said that the Skill can identify outdated explanations and temporary workarounds, and recommends enforcing constraints through Lint rules or the type system.
Necessary explanations may still be retained for external constraints that cannot be resolved independently, such as browser compatibility and third-party dependencies; public API documentation and license notices are also exceptions.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

This recommendation holds that certain comments pass past temporary practices on to subsequent AI. Rather than relying on comments as reminders, executable requirements should be written into Lint rules or the type system so that tools directly check for incorrect implementations.

Why it matters to readers

If this approach is adopted, teams may place greater emphasis on automatically enforceable code constraints, reducing Coding Agent the likelihood of repeatedly reusing outdated patches.

Beginners can divide comments into categories: explanations of external constraints that remain valid, or records of requirements that should be directly guaranteed by code rules. The latter are better converted into automated checks.

Risks and unknowns

The report does not provide a link to Lauren Tan's original interview or direct remarks.
The evidence alone cannot confirm that Dune indeed prohibits all code comments.
The evidence alone cannot confirm the actual code or usage of pstack and the `/no-comments` Skill.
“AI Generating comments contains 99%"explaining the historical background" is Tan's estimate and does not represent independent verification.

Grok Bot engineering lead calls for deleting the vast majority of code comments, saying they AI solidify temporary patches

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center