Skills: Teach AI how it can look at the output#6971
Conversation
Frame renders via the CLI can help verify visual output, but should remain an agent-chosen step rather than an implied default workflow. Made-with: Cursor
|
Reviewed PR #6971 — no actionable issues. Approved both the initial commit and the follow-up editorial simplification. The change adds a well-scoped optional sanity-check section to the Remotion skill file.
|
There was a problem hiding this comment.
Clean and well-scoped. The "optional" framing is explicit, npx usage is consistent with the rest of the skill files, and the --frame/--scale defaults are sensible for a quick sanity check.
Big Pickle (free) ｜ 𝕏
There was a problem hiding this comment.
New commit simplifies the paragraph into two clear sentences. The "Optional" heading still conveys the intent. LGTM.
Big Pickle (free) ｜ 𝕏
Some projects extend the
remotion-best-practicesskill with aremotion stillstep so agents double-check visual output. That reads like a mandatory loop when it sits in the main flow without qualification.This PR documents the same technique in upstream, but explicitly as an optional check the agent chooses when it helps—not part of the default workflow.
Relates to teams (e.g. birthday-bash) that found the unqualified wording too heavy-handed.
Made with Cursor