Track AI-isms / AI-slop tells across this repository's writing and any UI/visual work.
Tracking issue: encryption4all/dobby#44
What to check
Prose (READMEs, in-repo docs, code comments, PR/issue text, error messages, UI copy, log strings) against the 72 writing patterns to avoid: see process/writing-ai-slop-tells.md in the dobby memory repo.
Visual / UI work (if this repo ships UI) against the 48 design patterns to avoid: see process/design-ai-slop-tells.md in the dobby memory repo.
What the agent should do
- Walk the repo. Grep for the high-signal vocabulary tells (delve, leverage, robust, comprehensive, seamless, navigate, harness, foster, unlock, realm, landscape, tapestry, pivotal, transformative, plethora, myriad, etc.) and the phrase tells ("In today's...", "In the ever-evolving...", "It's worth noting...", "Embark on a journey", "Not just X — but Y", em-dash overuse, "Conclusion" / "Key Takeaways" sections, rule-of-three lists, etc.).
- Hand-review hits. Distinguish real AI-slop from words that happen to appear in technical context (e.g. "robust" in a security sense may be appropriate; "leverage" rarely is).
- For UI repos: also screenshot the running app and compare against the design tells (purple/violet primary, generic shadcn defaults, drifting spacing, missing focus states, etc.).
- Open one PR per concerned area. Keep changes mechanical where possible. Flag ambiguous calls in the PR description for human review. Do not invent new copy if the original meaning is unclear — ask.
- Reference this issue from each PR ("Refs #N").
To start the work
A human should comment on this issue with:
/dobby
A fresh agent will then pick it up and operate against this repo only. (The bot's webhook ignores comments from its own account, so the spawn comment must come from a human.)
Track AI-isms / AI-slop tells across this repository's writing and any UI/visual work.
Tracking issue: encryption4all/dobby#44
What to check
Prose (READMEs, in-repo docs, code comments, PR/issue text, error messages, UI copy, log strings) against the 72 writing patterns to avoid: see
process/writing-ai-slop-tells.mdin the dobby memory repo.Visual / UI work (if this repo ships UI) against the 48 design patterns to avoid: see
process/design-ai-slop-tells.mdin the dobby memory repo.What the agent should do
To start the work
A human should comment on this issue with:
/dobby
A fresh agent will then pick it up and operate against this repo only. (The bot's webhook ignores comments from its own account, so the spawn comment must come from a human.)