Introduces a plan/execute mode system for the AI agent. In plan mode the
agent is read-only — it analyzes footage, asks clarifying questions, and
submits a structured step-by-step plan. The user reviews and approves
before the agent switches to execute mode to apply edits.
New tools: submit_plan, ask_user, request_plan_approval, update_plan_step.
UI: plan panel checklist, mode transition modal, question prompt with
options, and Enter/Shift+Enter mode toggle in chat input.
Adds a skill discovery and loading system that lets the LLM agent
autonomously find and follow pre-built editing recipes. Skills are
exposed as two new tools: list_skills (search/discover) and load_skill
(get full instructions).
Built-in skills:
- viral-short: high-retention TikTok/Reels/Shorts workflow with hook,
retention cuts, zoom effects, captions, and CTA
- pitch-video: professional pitch/explainer with 5-act structure
(opening, problem, solution, metrics, CTA)
Luis Esteban Acevedo Ladino