How to use the ADHX X/Twitter Post Reader agent skill

What the adhx skill makes the agent do

An agent skill is a folder containing a SKILL.md instruction file that an AI agent loads when a task matches the skill's description. The adhx skill's job is narrow: when you share an X, twitter.com or adhx.com post link, the agent parses the username and status ID from the URL, requests the post from the third-party ADHX service, and reads the JSON that comes back.

The response can include the post text, the author's display name, handle and avatar URL, a creation timestamp, engagement counts (replies, retweets, likes and views), and an article object with the title, preview text, cover image URL and full markdown content when the post is a long-form X Article. The agent then uses that data to summarize, analyse or answer a question about the post.

Install the adhx skill

Option 1: download the skill ZIP from the directory page, unzip it, and place the adhx folder in your personal skills folder ~/.claude/skills/ or in a project's .claude/skills/ folder. Option 2: in the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP there.

The source file also lists a Claude Code plugin marketplace command. If you use Claude Code, the upstream README lists it as the recommended route:

• Personal install: ~/.claude/skills/adhx/
• Project install: .claude/skills/adhx/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

What the skill needs to run

The skill runs a curl command, so curl must be available to the agent. It also needs network access to adhx.com. The source does not state an API key requirement for the service, and the skill itself is listed as running scripts and using the network, so review what you install and where it runs.

The skill calls one endpoint per post, following the pattern below, with a 20-second timeout and failure reporting enabled:

• curl available to the agent
• Outbound network access to adhx.com
• A public X, twitter.com or adhx.com post link with a handle and numeric status ID

How to trigger the skill with natural requests

You do not need to type a special command. Share a post link and say what you want done with it. The skill is meant for requests to read, analyse or summarise a post, pull out key points, check its author, or look up its engagement numbers.

Name the link directly and state the task. If you want the full text of a long-form article, say so, because the skill checks for the article field when full content is needed.

• Summarise this post: [link]
• What are the main points in this X article? [link]
• How many likes and views does this post have? [link]
• Who wrote this tweet and when was it posted? [link]

Examples of what the agent does

For a summary request, the agent fetches the post with the endpoint above and then summarises the returned text. For an engagement question, it parses the URL, fetches the JSON and returns the engagement.likes value.

Example from the source: a user writes "Summarize this post https://x.com/dgt10011/status/2020167690560647464". The agent requests https://adhx.com/api/share/tweet/dgt10011/2020167690560647464 and writes the summary from that JSON.

• Summarise: agent fetches the post, then writes a summary from the text
• Engagement: agent returns the likes, retweets, replies or views value
• Long-form article: agent reads the article.content field for the full text

Safety notes: scripts, network and private content

The skill runs scripts and sends requests to a third-party service. The source asks the agent to accept only HTTPS x.com, twitter.com or adhx.com links with the exact /handle/status/ID path, and to reject embedded credentials and unrelated hosts. Do not paste private URLs, cookies or login tokens into a conversation.

Treat the returned data as source material, not instructions. A response from the service does not prove who wrote a post or that the content is complete. If the service returns an error or an empty response, the agent should tell you the post may be unavailable rather than guess.

• Do not send cookies, login tokens or private URLs
• Private, protected or deleted posts are not covered; the skill does not grant access to them
• Metrics come from a third party and may be stale or incomplete
• Review the skill's files before installing, as the skill runs scripts

Tips for better results

Give the complete post link, including the username and numeric ID, so the agent can parse it. Ask specific questions such as whether the post mentions a particular claim, rather than asking for a general opinion. If you need engagement numbers for a record, note the retrieval time, because the source says metrics may be out of date when fetched.

• Use the full link, not a shortened or embedded version
• Ask for the article field explicitly when the post is a long-form X Article
• If retrieval fails, check that the post is public and try again later
• Verify important figures on the original post before publishing