forge-agent-prompt

What problem does it solve?

It prevents orchestrators from wasting tokens and causing misbehavior by giving subagents briefs that are missing return shapes, autonomy boundaries, or scoped context.

Core Features & Use Cases

• Four-part brief structure: goal, context, constraints, and return shape in that exact order.
• Guardrails against prompt failure patterns: rejects compound goals, motivational filler, missing autonomy/budget, and vague “investigate X” instructions.
• Ready-to-paste templates: investigation, implementation, and verification brief patterns to reduce iteration time.
• Concrete return formats: JSON or labeled markdown with explicit length caps and “what not to include” rules.

Quick Start

Use the forge-agent-prompt skill to generate a self-contained subagent brief that includes a one-sentence goal, scoped context, explicit autonomy and budget, and a concrete JSON return shape.