plan-reviewer-agent

Review written plans for ambiguity, hidden assumptions, and handoff readiness.

2|Updated Apr 29, 2026
One-click install
npx skills add https://github.com/siarhei-belavus/agent-public --skill plan-reviewer-agent-siarhei-belavus
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: plan-reviewer-agent
Source: https://github.com/siarhei-belavus/agent-public/tree/main/skills/review/plan-reviewer-agent
Command: npx skills add https://github.com/siarhei-belavus/agent-public --skill plan-reviewer-agent-siarhei-belavus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Plans handed to another agent often contain vague wording, unresolved choices, and missing context that force the executor to guess. This Skill reviews a plan as an execution contract and determines whether it is safe to hand to a fresh agent with no prior conversation. ## Core Features & Use Cases - Ambiguity and Self-Sufficiency Passes: Flags vague terms, undefined targets, hidden prerequisites, and steps that require context the plan does not include. - No-Choice-Language Enforcement: Detects "or", "if needed", "as appropriate", and similar optionality inside execution instructions. - Structured Two-Part Output: Produces a compact human-readable verdict table plus a machine-friendly PLAN_FIX_PACKET with blocking issues and required rewrites. - Use Case: Before handing an implementation plan to a coding agent in a clean context window, run this review to confirm the plan is unambiguous, self-contained, and directive. ## Quick Start Review this plan for ambiguity, missing context, and handoff readiness before I send it to another agent.

Frequently Asked Questions about plan-reviewer-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if a plan is ready to hand off to another AI agent?▼

Review the plan as an execution contract: it must be unambiguous, self-contained, directive, and explicit about validation. This Skill simulates a fresh agent receiving only the plan and marks it not ready if any step requires guessing or follow-up questions.

What makes a plan ambiguous for agent execution?▼

Ambiguity comes from vague verbs like "handle" or "improve", unnamed files or owners, pronouns without clear referents, and choice-language like "or" or "if needed" inside execution steps. The review flags each occurrence with concrete rewrite instructions.

Can this review non-software plans like operations or research plans?▼

Yes, the review standard is project-agnostic and applies to software, operations, documentation, migration, and research plans. It requires equivalent clarity, validation criteria, and handoff completeness without forcing software-specific sections.

What output does the plan review produce?▼

It produces two parts: a compact single-column Markdown table with the verdict and severity-ordered findings for humans, and a PLAN_FIX_PACKET block with blocking issues, required changes, and items to preserve for the plan's author.

When should a plan be marked not ready?▼

A plan is not ready when any important term is undefined, any step requires context from prior conversation, execution instructions contain unresolved options, or validation and stop conditions are missing. Even one material guess blocks approval.