run-eval-loop

A conversational discovery skill that transforms vague ideas into clear specifications through adaptive interviews, auto-calibrating between quick scoping (3-5 questions) and deep multi-round discussions based on complexity, with optional spec.md or scope contract output.

Decomposes specifications or architecture documents into buildable tasks with acceptance criteria, dependencies, and implementation order, producing a structured tasks.md file for AI agents or engineers.

Multi-agent discussion rooms that debate or poll a problem from multiple perspectives, with debate mode for convergent argumentation in rounds and poll mode for independent analysis with consensus aggregation.

A composable stack of 39 AI agent skills spanning research, marketing, product development, and meta workflows. Skills chain together, with each reading artifacts left by earlier ones, so output compounds across the stack.

Part of the forsvn-skills collection of 39 composable skills that turn an AI agent into a full product team, covering research, marketing, product development, and meta workflows that chain together for compounding output.

Skill

Skill

Skill

Skill

Skill