read

name	read
description	Read a tweet from X (Twitter) by URL or ID and load it into the conversation. Use when user says "/x:read <url>", "read this tweet", "what does this tweet say", or pastes an x.com / twitter.com status link and wants to discuss, summarize, or reply to it. NOT for posting (use /x:tweet), timelines, search, analytics, or DMs.
argument-hint	<x.com URL or numeric tweet ID> [--json]
allowed-tools	Bash(python3*), Read
user-invocable	false

Read - Fetch a Tweet via X API

Fetch a single tweet via the X API v2 GET /2/tweets/:id endpoint and render it as markdown so the rest of the conversation can act on it (summarize, reply via /x:tweet --reply-to, quote, etc.).

Workflow

1. Parse target: From $ARGUMENTS, extract the URL or numeric tweet ID as a single positional argument, and capture any --json (or natural-language equivalent like "raw" / "as JSON") as a separate flag. Accepted target forms: https://x.com/<user>/status/<id>, twitter.com/..., or a bare numeric ID — the script handles all three.
2. Fetch: From this skill's directory, run python3 scripts/read.py <target>, appending --json as a separate argument when raw output was requested. Do NOT pass "$ARGUMENTS" as a single quoted string — that swallows flags. The script reuses the same Keychain credentials as /x:tweet (service x-api).
3. Present: The script prints markdown with the tweet's text, author, timestamp, public metrics, any referenced tweets (replied-to / quoted), and media URLs. Show this output to the user verbatim or summarize it depending on what they asked for.
4. Follow-up: If the user wants to reply, hand off to /x:tweet with the tweet ID for --reply-to. The fetched tweet's URL contains the ID after /status/.

Long-form articles

When a tweet is an X long-form article (the body is a t.co link wrapping x.com/i/article/...), the script:

• Renders the header as # Article by ... instead of # Tweet by ...
• Surfaces the article title under ## Article: <title>
• Adds an **Article URL:** line pointing at the expanded x.com/i/article/:id
• Emits a clear fallback note that the API exposes only the title, not the body — the user must open the URL to read the full piece. If data.note_tweet.text is populated by the API (rare), that body is rendered inline.
• Still surfaces the wrapping tweet's own text under ### Wrapping tweet text

If the user needs the body content for downstream work, hand off to a browser tool (e.g., mcp__chrome-devtools__*) to navigate the article URL and extract — /x:read itself stops at the API layer.

Flags

• --json — print the raw API payload instead of markdown. Use when the user asks for raw data, full entities, or fields the markdown view drops.

Errors

• Missing credentials: Direct user to run /x:tweet once (it triggers scripts/setup.sh for native macOS keychain dialogs), or set X_API_KEY / X_API_SECRET / X_ACCESS_TOKEN / X_ACCESS_TOKEN_SECRET env vars.
• 404: Tweet was deleted, is from a private account, or never existed. Don't retry.
• 429: Rate limited. Wait and try again — don't loop.
• 401/403: Credentials are invalid or the account lacks read access for that tweet. Re-run setup.

Cost

X API v2 Posts read tier is billed per resource (~$0.005 / post read + ~$0.010 / user read on the standard tier as of 2026). One /x:read call = one post read + (typically) one user read for the author.