Tessl Proposes Attributed Agent Mistake Logs for Team Knowledge

Tessl's blog post argues that teams should record agent mistakes as attributed, signed diary entries and curate them into reusable context packs, rather than adding unverified rules to files such as AGENTS.md.

The author illustrates the problem with an agent that regenerated an OpenAPI spec and TypeScript client but missed the Go client, and a lesson that had to be re-taught in a fresh session. The post presents this as a proposed practice, and its claims about effectiveness are the author's own; the report does not independently evidence them.

A separate X post by Lauren Tan (@poteto) takes a different route: she says codebases act as agent memory, and that each recurring agent mistake should become a lint rule, type constraint, or refactor that makes the anti-pattern impossible, and that agents should be given a way to verify their own work.