Removing AI Tells from Your Writing: A Skill That Turns One Flag into a Permanent Rule
The posts on this site start as agent drafts that I edit. The drafts come with the awkward phrasing AI is known for. At first I fixed each case by hand, and the next post carried the same disease. I was fixing sentences, not habits. So I changed the setup. I no longer fix awkward sentences. I only flag them. The agent takes the flag, names the pattern, and stores it as a rule in a skill file.…
The article discusses a method to identify and eliminate artificial intelligence (AI) language patterns from written drafts. The author, who uses AI-generated drafts, initially fixed each case manually but soon realized that this only addressed the symptoms and not the underlying habits. To address this, the author changed the approach: instead of fixing the sentences, the drafts now have flags that identify the patterns, which are then recorded as rules in a skill file.
This allows the AI agent to apply these rules to future drafts, turning a one-time edit into a permanent system.
The skill file consists of four parts: a procedure, a list of AI language patterns (lets), another list of English lets, and the voice the author wants to maintain. The procedure involves reading the entire draft, fixing rule violations, checking the rhythm by reading aloud, running a community skill as a second pass, and reporting changes.
The author emphasizes that the rules are not just about grammar but about habits, and that deleting a sentence before repairing it can sometimes be more effective. The file also includes a list of banned words and specific grammatical structures to avoid.
The author provides a detailed example of how a rule was born from a specific sentence in a draft. The sentence "The door opens first where being wrong is affordable" was initially flagged with a simple "this phrasing is off" comment. However, the author then renamed it to "a metaphor doing the work of a claim, with no real subject," revealing that the issue was not just about translation but about a habit of using metaphors inappropriately.
This rule targets the habit rather than the language directly, making it applicable to all languages.
After establishing the initial set of rules, the author adds a follow-up rule that suggests deleting the sentence if it trips a rule and the meaning can survive. This principle has been incorporated into the procedure as the second step. The author concludes that while sentence-level rules can't catch everything, focusing on the system of rules and the underlying habits leads to cleaner drafts over time.
Even though some AI tells remain, the author publishes the writing anyway, viewing the process as a record of their learning journey.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written; read the original for the full account.
