x-reader

# X/Twitter Post Reader Read tweet content and replies without Twitter API credentials. ## URL Parsing Extract `{username}` and `{statusId}` from the URL path segments. Supported formats: - `https://x.com/{user}/status/{id}` - `https://twitter.com/{user}/status/{id}` - `https://adhx.com/{user}/status/{id}` - `https://x.com/{user}/article/{id}` (long-form articles — use statusId from URL) ## Step 1: Fetch Tweet Content (ADHX API) ```bash curl -s "https://adhx.com/api/share/tweet/{username}/{statusId}" ``` No auth required. Returns JSON with these fields: - `text` — tweet body (empty string for article-only posts) - `author` — `{name, username, avatarUrl}` - `engagement` — `{replies, retweets, likes, views}` - `article` — present ONLY for long-form X Articles: - `article.title` — article title - `article.content` — full article body as Markdown (includes `` image references) - `article.coverImageUrl` — hero image **Handling articles vs tweets:** - If `article` field exists → the post is a long-form article. Use `article.title` and `article.content` for the body. - If `article` field is absent → regular tweet. Use `text` for the body. **Fallback** if ADHX is down: ```bash curl -s "https://api.fxtwitter.com/{username}/status/{statusId}" ``` ## Step 2: Fetch Replies (DevTools Network Interception) Replies are not available via any free API. Instead, intercept X.com's own GraphQL request: 1. **Find or open the X.com tab** via chrome-devtools-mcp: - `list_pages` to check if the tweet is already open - `select_page` to select the X.com tab, or `navigate_page` to open the tweet URL 2. **Navigate to the tweet page**: ``` navigate_page(type="url", url="https://x.com/{username}/status/{statusId}") ``` 3. **List network requests** filtered to fetch/xhr: ``` list_network_requests(resourceTypes=["fetch", "xhr"]) ``` 4. **Find the TweetDetail request** — look for a request URL containing `/graphql/` and `TweetDetail` with `focalTweetId` matching the statusId. 5. **Save the response** to a temp file: ``` get_network_request(reqid=<id>, responseFilePath="D:/Agent/tmp/tweet-replies-{statusId}.json") ``` 6. **Parse replies** using the bundled script: ```bash python3 ${CLAUDE_SKILL_DIR}/scripts/parse_replies.py "D:/Agent/tmp/tweet-replies-{statusId}.json" ``` The script returns JSON: ```json { "replies": [ {"author": "...", "username": "...", "text": "...", "likes": 123, ...} ], "cursor": "pagination_cursor_or_null" } ``` ## Step 3: Present Results Combine tweet content and replies into a readable summary: 1. **Tweet header**: author, date, engagement stats 2. **Tweet body**: full text (summarize if user asks) 3. **Top replies**: sorted by likes, show author + text + likes count ## Fallback If chrome-devtools-mcp is unavailable (no Chrome open, no DevTools connection), skip Step 2 and inform the user that only the tweet body is available without replies. ## Notes - ADHX API is free, no auth, no rate limit (be polite) - DevTools interception uses the user's logged-in Chrome session - Each page load returns ~30 top replies (ranked by relevance) - Clean up temp files in `D:/Agent/tmp/` after parsing