Research before committing to the brief
blindspot-pass makes an agent pause before implementation and inspect unfamiliar territory. It reports likely landmines, hidden constraints, examples of good work, and questions an expert would ask. It then rewrites the original request with that context included.
Use it before a costly change in a part of the codebase or domain you do not know well. Give it the intended result and your experience level, then check that every addition to the rewritten brief has evidence behind it.
The skill deliberately stops before doing the task. That makes it unnecessary overhead for small, familiar work, and its external research is only as good as the sources the host agent can reach. The final brief still needs a human to separate real constraints from generic caution.