commentmd lets an AI agent open a Markdown review session in the browser, collect structured human comments, and revise the document with precision.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install md-comment
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install md-comment using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is commentmd?
commentmd is an Openclaw Skills workflow for turning Markdown review into a guided, browser-based feedback loop. When an agent has generated a design doc, tech spec, README, or PR description, it can launch a local review page so a human can highlight text, leave targeted comments, and submit structured feedback the agent can parse.
This skill is designed to reduce back-and-forth, preserve comment context with exact text anchors, and keep review artifacts machine-readable. Openclaw Skills users can use commentmd to safely insert human approval into an AI writing workflow without losing traceability or forcing freeform feedback.
commentmd Use Cases
- Reviewing AI-generated Markdown documents before finalizing them.
- Collecting precise feedback on design docs, technical specs, READMEs, and PR descriptions.
- Letting a human annotate exact text spans instead of leaving vague general comments.
- Running a review session on a local machine with automatic browser launch.
- Supporting remote or headless workflows where the reviewer downloads a static HTML review page.
- Feeding structured review output back into the agent so it can revise the source document intelligently.
How commentmd Works
- The agent receives a Markdown file path from the user or detects that a generated document needs human review.
- It resolves the Markdown path to an absolute file location.
- It starts the review server with the provided Python script, which serves the document locally and opens a browser tab.
- The reviewer highlights text, adds comments, and clicks the Finish button to submit feedback.
- The server writes a structured JSON file next to the source Markdown file, usually named
<md_name_without_ext>.comments.json. - The agent reads the JSON output, inspects each comment's quote and text anchors, and checks whether the document changed during review.
- The agent revises the original Markdown file and summarizes how each comment was addressed.
- In remote or headless environments, the agent serves a static review page instead and ingests the downloaded JSON file after review.
commentmd Setup
- Ensure Python 3 is available in the environment.
- Make sure the skill files are installed under
~/.agents/skills/commentmd/. - Review a Markdown file by passing its path to the skill command:
/commentmd <md_path>
- The agent will resolve the path and run the review server:
python3 ~/.agents/skills/commentmd/scripts/serve.py "$ABS"
- For headless or remote environments, generate a static review page instead:
python3 ~/.agents/skills/commentmd/scripts/serve.py "$ABS" --static /tmp/review.html
- Open the served page, leave comments, click Finish, then provide the downloaded JSON path if the review was completed remotely.
- After the JSON is produced, read it and update the source Markdown accordingly.
commentmd Data Schema & Taxonomy
commentmd produces a structured review artifact that keeps Markdown feedback machine-readable and easy to apply.
| Field | Type | Purpose |
|---|---|---|
schema_version |
number | Version of the review JSON schema. |
md_file |
string | Absolute path to the reviewed Markdown file. |
md_sha256 |
string | Hash of the reviewed Markdown content for integrity tracking. |
md_changed_during_review |
boolean | Indicates whether the source file changed while the review was open. |
created_at |
string | ISO-8601 timestamp for when the review was created. |
comment_count |
number | Total number of comments submitted. |
comments |
array | Ordered list of structured review comments. |
Each object in comments contains:
id: stable comment identifier such asc1.quote: the exact excerpt selected by the reviewer.prefix: a 32-character anchor window before the quote.suffix: a 32-character anchor window after the quote.comment: the human review note or requested change.created_at: timestamp for the individual comment.
Output behavior:
- The default output file is written next to the source Markdown file as
<md_name_without_ext>.comments.json. - A
comment_countof0means the reviewer had no objections. - If
md_changed_during_reviewistrue, the agent should warn the user that the source changed externally during the review session.
commentmd Advanced Features
- Opens a local HTTP server on
127.0.0.1:3118, automatically moving to the next free port up to3128if needed. - Supports browser-based highlight-and-comment review for precise text anchoring.
- Generates structured JSON output so the agent can process feedback deterministically.
- Detects external file changes during review with
md_changed_during_review. - Supports remote and headless workflows via
--static /tmp/review.html. - Preserves exact quote context with
prefixandsuffixanchor windows for reliable comment mapping. - Works especially well in Openclaw Skills pipelines where human approval is needed before the agent proceeds.
SKILL.md
Loading
Related Openclaw Skills
Break Watch
A Python-powered A-share stock market scanner designed to identify real-time breakout patterns combining volume surges, upward price movements, and bullish candles.
Browser Search
A multi-layered web search and browsing engine that scales from lightweight SearXNG queries to advanced CloakBrowser anti-bot evasion.
Drinks Sommelier
An intelligent, image-enabled AI sommelier that evaluates beer and wine selections against a dynamically updated profile of your personal taste preferences.
Makaron CLI Creative Agent
A CLI tool and agent skill to dynamically generate, edit, and assemble images, video, music, and motion designs.
commentmd
commentmd turns a Markdown file into a browser-based review session that returns structured human feedback the agent can act on.
commentmd
commentmd turns any Markdown file into a browser-based review session that returns structured human comments the agent can act on.
Featured*
