Plan Goal Checker

Overview

This agent performs critical, goal-backward verification on proposed project plans. It does not execute the plan; instead, it rigorously analyzes the plan to ensure that every step logically contributes to achieving the defined phase goal, preventing wasted context and failed execution attempts.

It acts as a safety net, ensuring that the intent described in the plan matches the deliverable outcome required by the project goals.

Capabilities

• Goal-Backward Analysis: Starts from the desired final output (the phase goal) and traces backward to verify if all planned tasks logically support it.
• Contradiction Detection: Checks plans against explicit user decisions documented in CONTEXT.md to flag any discrepancies.
• Dependency Mapping: Identifies potential broken dependencies, circular references, or missing artifact wiring between planned steps.
• Contextual Compliance Check: Verifies that the plan adheres to project-specific guidelines found in ./CLAUDE.md and accounts for available project skills.

Example Use Cases

• Pre-Execution Gatekeeping: Before running a multi-step data processing pipeline, use this agent to confirm all necessary inputs are accounted for and that the final output meets the required schema.
• Plan Revision Review: After an initial plan draft is created, run this checker before committing resources to ensure revisions have closed all identified gaps.
• Scope Validation: When a project scope seems too large or vague, use this agent to force a breakdown and verify that each proposed sub-task directly maps back to a critical requirement.