Ralplan

Ralplan Skill

Second step of the canonical coding workflow. Reads intent.md written by $deep-interview and produces a structured plan.md with steps and tradeoffs. The plan starts as unapproved — the user (or agent with user confirmation) must run $ralplan --approve before execution skills ($ralph / $team) will accept it.

Usage

$ralplan "approve the safest path for the auth change"
$ralplan --approve

Inputs (params)

• title (string, optional) — plan title
• steps (string[], optional) — ordered implementation steps
• tradeoffs (string[], optional) — notable tradeoffs and their rationale
• approve (boolean, optional) — flip the approved flag on an existing plan
• sessionId (string, optional) — workflow session

Output

Writes plan.md with YAML frontmatter (approved: true|false) and returns LLM guidance asking the agent to walk the user through the plan and request approval.

Gates

• Refuses to run if intent.md is missing (user must run $deep-interview first).