Offline-searchable X/Twitter CLI and MCP surface for archiving posts, resolving links, monitoring mentions, composing threads, publishing Articles, and searching synced bookmarks. Trigger phrases: `search X for`, `archive tweets about`, `show me the X thread for`, `monitor my X mentions`, `post a th

<! GENERATED FILE — DO NOT EDIT. This file is a verbatim mirror of library/social and messaging/x twitter/SKILL.md, regenerated post merge by tools/generate skills/. Hand edits here are silently overwritten on the next regen. Edit the library/ source instead. See the repository agent guide, section "Generated artifacts: registry.json, cli skills/". X (Twitter) — Printing Press CLI Prerequisites: Install the CLI This skill drives the x twitter pp cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first: 1. Install via the Printing Press installer. It defaults binaries to $HOME/.local/bin on macOS/Linux and %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows: 2. Verify: x twitter pp cli version 3. Ensure the reported install directory is on $PATH for the agent/runtime that will invoke this skill. If the npx install fails (no Node, offline, etc.), fall back to a direct Go install (requires Go 1.26.6 or newer). This installs into $GOPATH/bin (default $HOME/go/bin ), so add that directory to $PATH instead: If version reports "command not found" after install, the runtime cannot see the binary directory on $PATH . Do not proceed with skill commands until verification succeeds. Mirrors the official X v2 API and adds what no other X tool has: a local SQLite store you can sync once and query many times with FTS5 search and group by analytics, agent native json/ select output, and an MCP server that exposes the whole surface to AI agents through token efficient orchestration plus named multi step intents. Start pasted link workflows with post resolve , use thread context when a post needs parent/quote/reply context, save durable source material with collection save/list/export , track ongoing searches with monitor create/run/list , package saved activity with brief , snapshot accounts with account snapshot , find launch or repo links with url mentions , track post metrics with performance snapshot/backfill/analyze , export account/query timelines with timeline export , reconstruct synced conversation threads offline with thread show , compose self reply threads from markdown with thread compose , author long form X Articles from markdown with articles publish md , and rescue your bookmark graveyard with users bookmarks find — keyword and author search over your synced bookmarks, which X itself gives you no way to search. When to Use This CLI For a ranked readout of recent posts, run x twitter pp cli top posts metric engagement limit 10 max fetch 100 json . It reads the live X timeline, so each page may use paid API credits. max fetch accepts 1 to 1000 posts, with a small extra page budget for sparse responses. A leaderboard returned at that budget is partial and has truncated: true in every JSON row. user id selects a numeric account ID without a separate username lookup; without it, the command uses the authenticated user. JSON rows name the effective score metric , and missing impression scores are null rather than zero. dry run makes no request. Reach for this CLI when a task involves reading, searching, or archiving X (Twitter) data and you want the results queryable offline rather than re fetched each time — building a searchable corpus of posts, reconstructing a conversation thread, snapshotting a user's recent posts with engagement, or monitoring mentions incrementally. It is also the right choice when an AI agent needs an X surface with read only/destructive safety hints and named multi step intents rather than a pile of raw endpoint calls. Prefer it over raw API calls whenever the same data will be queried more than once, since the local store avoids re spending per read credits. Unique Capabilities These capabilities aren't available in any other tool for this API. Local state that compounds post resolve — Normalize any X post URL or raw post ID into a canonical structured record. It prefers local data in auto mode, falls back to public v2 reads when needed, includes provenance ( live , local , or mixed ), and emits suggested next workflow commands. Use this as the first command when an agent starts from a pasted X link. It avoids brittle URL parsing and returns stable fields for downstream workflows. thread context — Resolve a post URL or ID, include parent and quoted posts when available, and optionally include bounded replies from the local store and/or recent search. Use this before summarizing or drafting around a post. thread show is still the pure offline conversation reconstruction command; thread context is the URL first workflow that can mix local and live data. collection save/list/export — Save resolved X posts into durable named local collections, then list or export them as markdown, JSON, JSONL, or CSV. This never writes to X; it only writes the local SQLite store. Use collections for research libraries, launch mentions, examples, recruiting/source material, and anything an agent should reuse offline after the first API read. monitor create/run/list — Create named query, URL, or account monitors, then run them repeatedly with local watermarks and dedupe. monitor create and monitor run write only local SQLite state; they never post, like, reply, follow, or mutate X. Use monitors for launch mentions, product/customer feedback, account tracking, and recurring agent jobs. since last uses the saved watermark; preview fetches without updating local state. brief — Build a deterministic source backed JSON or markdown brief from monitor results, collections, or explicit post IDs. It packages links, authors, text, and available metrics; it does not invent conclusions or run LLM summarization. Use this when an agent needs a daily/weekly X activity packet or a collection summary that can be pasted into notes, docs, or a CRM. account snapshot — Capture profile basics, public metrics, pinned post, and recent posts for a username or user ID. It is read only toward X and uses local data first unless live or data source live is set. url mentions — Search for recent posts mentioning a URL, domain, repo, article, or product page. It can optionally save results into a local collection or create/update a local monitor for future runs. performance snapshot/backfill/analyze — Store timestamped post metrics locally, backfill recent account posts when auth allows, and analyze saved snapshots by type, hour, media, link presence, or label. Missing metrics stay nullable/absent; the CLI does not treat unavailable fields as zero. Snapshot output includes public metrics , available non public/organic metrics, metric source , and metric availability . timeline export — Export an account or query timeline as markdown, JSON, or JSONL. Account exports use local synced tweets when available and fetch live only when needed or requested. thread show — Rebuild a full conversation thread from your locally synced posts — ordered and depth tagged — without re spending API read credits. When an agent needs the shape of a discussion (who replied to whom, in order), reach for this instead of paginating the search API and re assembling the tree by hand. users bookmarks find — Search your synced bookmarks by keyword and/or author, offline. X has no bookmark search and its API exposes no bookmark timestamp, so bookmarks pile up unread; this rebuilds the missing retrieval layer from the local store. Read only and free to re run once synced — the X read credit is spent once at sync time, not per query. The "I bookmarked something about this and can't find it" rescue. Sync bookmarks once, then let an agent retrieve and act on them (summarize, draft a thread, cluster) — the offline store is the agent's working set, not a per query API spend. Authoring workflows thread compose — Split a markdown file into a numbered, 280 char packed self reply thread; prints by default and only posts with post. Compose a thread from a document deterministically; the dry run default lets an agent preview the exact tweets before any write. articles publish md — Parse a markdown file with YAML frontmatter into the Draft.js content state JSON X's Articles editor accepts; previews by default; draft creates a new draft, update edits an existing draft, post publishes publicly. The only programmatic way to author a long form X Article from a document; preview, draft, and update keep it private until you explicitly post. The converter supports paragraphs, / headers ( and deeper are downgraded to ), lists, blockquotes, fenced code blocks, tables, dividers, image lines ( ![alt](path) ), standalone tweet URLs (native embedded posts: put https://x.com/<user /status/<id alone on its own line), bold/italic inline styles (including inside blockquotes; backtick code spans become plain text because X Articles has no inline code style), and [text](url) inline links, which become real LINK entities (working hyperlinks) in the article. Lines that are HTML comments ( <! ... ) are stripped and never ship as article text. articles update md — Update an existing X Article draft in place from a markdown file: title (frontmatter), body, links, and new inline images, all in one command. Find the article id via articles list agent . Edit the markdown, re run one command. No duplicate drafts: the draft keeps its id, so the edit URL and any saved state stay valid. WARNING — body replacement: articles update md replaces the ENTIRE body of the target draft. Anything added by hand in the X composer that the markdown converter cannot reproduce is destroyed by the update. The command preflights the draft's content state and refuses when it finds entity types outside the converter's emit set (LINK, MARKDOWN, MEDIA, TWEET, DIVIDER); pass replace unknown entities only when you genuinely want to overwrite those artifacts. Markdown authored content round trips cleanly; if a draft holds composer only embeds or other manual artifacts, edit it in the composer instead. Draft only by default: a published article is refused unless you pass republish , which runs Unpublish UpdateContent Publish (the article is briefly unpublished). Use dry run to preview the converted payload with no API call. articles patch md — Add sections, embedded posts or text fixes to an existing (even published) X Article without rewriting the rest. It reads the current body and keeps every block, image and embed as it is, then applies anchored edits from a small patch file. Use this instead of update md for an article that already has images or composer edits: nothing outside the patch changes, and no image is re uploaded. An anchor is the start of an existing paragraph or heading and must match exactly one block; a replace target must appear exactly once and may not cut across a link or bold span. Any miss stops the command before it writes. dry run still reads the article (it needs the real body) but never writes. Published articles need republish (the same Unpublish UpdateContent Publish sequence the composer uses; the post, its URL and its likes stay). If your cookies were captured without twid , set X TWITTER USER ID=<your numeric id . Command Reference account activity — Endpoints relating to retrieving, managing AAA subscriptions x twitter pp cli account activity create subscription — Creates an Account Activity subscription for the user and the given webhook. x twitter pp cli account activity delete subscription — Deletes an Account Activity subscription for the given webhook and user ID. x twitter pp cli account activity get subscription count — Retrieves a count of currently active Account Activity subscriptions.

pp-x-twitter