Skill Detail

Review and revise an agent-authored Markdown plan with Crit

Open a local Markdown plan for line-level human comments, return the feedback to the agent, and repeat the revision loop until the operator approves.

Code Quality & ReviewMulti-Framework
Code Quality & Review Multi-Framework Security Reviewed
⭐ 1.3k GitHub stars
COPY SKILL INSTRUCTIONS (OPTIONAL)
npx skills add agentskillexchange/skills --skill review-revise-agent-markdown-plan-with-crit Copy
Uses the third-party skills CLI, not an ASE-owned installer. Check your agent’s compatibility. This copies instructions; complete the upstream tool setup below separately.
At a glance
Tools required
Crit binary; local Markdown plan; agent able to read files and run commands; local browser for the operator
Install & setup
Install the reviewed Crit v0.23.2 binary from https://github.com/tomasz-tomczyk/crit/releases/tag/v0.23.2 for your operating system and place it on PATH. Run crit plan.md in the project. Use the upstream integration matching your agent, or manually return the review prompt to the agent after reviewing. No remote sharing or account is required for this local workflow.
Author
tomasz-tomczyk
Publisher
Community
Last updated
Nov 13, 2026
Quick brief

Use Crit when a coding agent has drafted a Markdown implementation plan and the operator needs to comment on exact lines before implementation begins. Supply the plan path, repository context and the decisions that need review. The output is a revised plan plus a record of addressed and unresolved comments.

How it works

What this skill actually does

Install the Crit binary, then run crit plan.md in the project to render the existing document in its local review interface. The operator reads the plan and attaches line or range comments. Return the review feedback to the coding agent using the documented integration or manual prompt handoff. The agent revises only the agreed plan scope, identifies any unresolved choices, and presents it for another review round. Stop when the operator explicitly approves the plan or requests more information.

Invoke this for a collaborative planning checkpoint when ordinary chat loses the location of comments or when a Git diff is premature. The workflow centers on the author’s existing Markdown document and targeted human feedback. It does not require converting the plan into an execution graph or opening a remote pull request.

Scope is one local plan and its comment-to-revision loop. Code implementation, autonomous approval, remote sharing, forge comment publishing and background agent execution are separate actions. Keep unresolved operator decisions visible; a completed command is not approval. Multiple runtimes can use the same file-and-command contract, so Multi-Framework is intentional.

Upstream v0.23.2 documents Markdown rendering, line comments and the agent review loop. Metadata is verified; runtime behavior and security have not been reviewed.