Skip to main content
Glama
hermoso-ai

Hermoso

Official

Publish a long-form Article to X

post_x_article
Destructive

Publish long-form X Articles from a title and Markdown body. Converts headings, lists, links, and tables into X's native format, refusing silently dropped formatting.

Instructions

Publish a long-form ARTICLE to the user’s connected X (Twitter) account — X’s own long-form format, which is a different thing from a long POST. Give it a title and a body written in MARKDOWN (or plain prose) and Hermoso converts it into the DraftJS content_state structure X requires: headings, paragraphs, bulleted and numbered lists, blockquotes, bold / italic / strikethrough, links, horizontal rules, fenced code blocks and pipe tables all carry across. FORMATTING IS NEVER SILENTLY DROPPED — anything X Articles cannot represent (inline code, an inline image) REFUSES the article for free and names exactly what and why, and allowLossy: true is the explicit way to publish it as plain text anyway. THE HARD LIMIT TO PLAN AROUND: X allows only about 10 Article DRAFTS and 5 Article PUBLISHES per account per DAY, it publishes that cap nowhere, and it counts a FAILED attempt against them — so never iterate on an article by republishing it, and use publish: false to save a draft for the user to read in X’s own composer when they want to review before it goes out. This PUBLISHES immediately and PUBLICLY: show the user the full text and get an explicit yes first, because a published Article can NEVER be edited (X’s own rule — edit_x_post will refuse it) and the only remedy is to delete and republish, which costs another of the five. An Article appears in the timeline as a title card, not as body text; readers open it. Costs credits (X bills per API request, and this is three of them). Needs X connected (Settings > Connectors > X).

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
bodyYesthe article body, as markdown or plain prose. Markdown headings, lists, quotes, links, emphasis, ``` code fences and | pipe | tables | are all converted to X’s own Article structure.
hookNoWHAT ANGLE this post is built on, recorded only at publish time. post_performance ranks hooks on it (a winner needs 5 posts sharing ONE hook), so pass a list_hooks id (e.g. "direct_callout", "before_after") or reuse your own wording EXACTLY across a campaign. Omit it and this post never votes on which hook works.
brandNoWHICH PROFILE this post belongs to — id or exact name from list_brands (a shared workspace: its profile id). Beats the connection's pin for THIS CALL ONLY; a name that matches no profile, or two, is REFUSED and nothing is posted.
titleYesthe Article title — X requires one and refuses a draft without it. This is what shows on the timeline card.
ideaIdNoshort id of the content-plan idea this post came from
recipeNothe post's FORMAT id, e.g. "slideshow" or "imessage_chat" — post_performance groups by it, so reuse one id per format
publishNodefault true. Pass false to save it as a DRAFT in the account’s X Articles composer instead — nothing becomes public, the user can review and publish it from X, and it does not spend one of the five daily publishes.
subjectNoWHAT THIS POST IS ABOUT — product, feature, offer or theme (e.g. "winter coat", "free trial"). post_performance's second grouping axis: reuse the exact wording, as with hook.
headingsNohow headings are rendered. “blocks” (default) uses X’s own heading block types for # and ## headings; ### and deeper become bold lines, because X Articles refuse a third-level heading (the reply says so). “text” renders every heading as a BOLD standalone paragraph instead: the words survive and the heading structure does not, and the reply says so.
allowLossyNopublish even though part of the source cannot be represented on X, rendering those parts as plain text. OFF by default and it should usually stay off — silently publishing a user’s copy with formatting missing is worse than refusing and telling them.
coverImageUrlNooptional cover picture for the Article — a Hermoso render URL or an upload_file url. Must be a STILL image; X Article covers are not videos.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changedv0.1.374
    • changedInput schema / properties / brand / description
      Previous value: -"WHICH BRAND this post belongs to — id or exact name from list_brands (a shared workspace: its profile id). Beats the connection's pin for THIS CALL ONLY; a name that matches no brand, or two, is REFUSED and nothing is posted."New value: +"WHICH PROFILE this post belongs to — id or exact name from list_brands (a shared workspace: its profile id). Beats the connection's pin for THIS CALL ONLY; a name that matches no profile, or two, is REFUSED and nothing is posted."
  2. Changed3 schema fields changedv0.1.320
    • changedInput schema / properties / brand / description
      Previous value: -"WHICH BRAND this post belongs to — the id or exact name from list_brands (a workspace shared with you: its profile id). Use it whenever the account has more than one brand and you are not certain which one this connection is pinned to: it beats the pin for THIS CALL ONLY and changes nothing about the connection. A name that matches no brand, or two brands, is REFUSED and nothing is posted — never resolved to the pin, which is the account you were guarding against."New value: +"WHICH BRAND this post belongs to — id or exact name from list_brands (a shared workspace: its profile id). Beats the connection's pin for THIS CALL ONLY; a name that matches no brand, or two, is REFUSED and nothing is posted."
    • changedInput schema / properties / hook / description
      Previous value: -"WHAT ANGLE THIS POST IS BUILT ON — the single most valuable field here, and the only moment it can ever be recorded. post_performance groups on it to answer \"which hooks work\", and it needs 5 posts sharing ONE hook before it will call anything a winner, so REUSE THE SAME WORDING across a campaign instead of rephrasing it every time. Best of all, pass a hook id from list_hooks (e.g. \"direct_callout\", \"mid_problem\", \"before_after\") — those fold onto a stable key however they are spelled, so a whole brand accumulates evidence on one row. Your own wording is fine too; it just only groups when you repeat it exactly. Omitting it means this post can never vote on which hook works."New value: +"WHAT ANGLE this post is built on, recorded only at publish time. post_performance ranks hooks on it (a winner needs 5 posts sharing ONE hook), so pass a list_hooks id (e.g. \"direct_callout\", \"before_after\") or reuse your own wording EXACTLY across a campaign. Omit it and this post never votes on which hook works."
    • changedInput schema / properties / subject / description
      Previous value: -"WHAT THIS POST IS ABOUT — the product, feature, offer or theme (e.g. \"winter coat\", \"free trial\", \"founder story\"). The second grouping axis in post_performance. Same rule as hook: reuse the exact wording so posts about one subject land in one group."New value: +"WHAT THIS POST IS ABOUT — product, feature, offer or theme (e.g. \"winter coat\", \"free trial\"). post_performance's second grouping axis: reuse the exact wording, as with hook."
  3. Changed2 schema fields changedv0.1.281
    • addedInput schema / properties / ideaId
      Added value: +{
      +  "description": "short id of the content-plan idea this post came from",
      +  "type": "string"
      +}
    • addedInput schema / properties / recipe
      Added value: +{
      +  "description": "the post's FORMAT id, e.g. \"slideshow\" or \"imessage_chat\" — post_performance groups by it, so reuse one id per format",
      +  "type": "string"
      +}
  4. Changed1 schema field changedv0.1.251
    • addedInput schema / properties / brand
      Added value: +{
      +  "description": "WHICH BRAND this post belongs to — the id or exact name from list_brands (a workspace shared with you: its profile id). Use it whenever the account has more than one brand and you are not certain which one this connection is pinned to: it beats the pin for THIS CALL ONLY and changes nothing about the connection. A name that matches no brand, or two brands, is REFUSED and nothing is posted — never resolved to the pin, which is the account you were guarding against.",
      +  "type": "string"
      +}
  5. Changed1 schema field changedv0.1.230
    • changedInput schema / properties / headings / description
      Previous value: -"how headings are rendered. “blocks” (default) uses X’s own heading block types, which is the faithful conversion. “text” renders each heading as a BOLD standalone paragraph instead: the words survive and the heading structure does not, the reply says so, and it is the documented fallback if X’s Articles service rejects heading blocks."New value: +"how headings are rendered. “blocks” (default) uses X’s own heading block types for # and ## headings; ### and deeper become bold lines, because X Articles refuse a third-level heading (the reply says so). “text” renders every heading as a BOLD standalone paragraph instead: the words survive and the heading structure does not, and the reply says so."
  6. Addedv0.1.161

TDQS

A4.5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Far exceeds annotations (destructiveHint=true, openWorldHint=true). Discloses uneditable-after-publish rule (edit_x_post refuses), daily draft/publish caps (~10/5) that are undocumented by X, failed attempts counting against the cap, credit cost (three API requests), lossy-format refusal behavior, and connector prerequisite. This is exactly the behavioral context annotations cannot carry.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Rich and front-loaded, but heavily oversized with ALL-CAPS emphasis and multi-clause sentences that pile on caveats. The formatting-conversion list and cap warnings are valuable, yet the draft reads more like a policy page than a tool description, and some redundancy (X's rules restated twice) could be trimmed.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a destructive, non-idempotent, open-world publish tool with no output schema, the description covers approval workflow, irreversibility, caps, cost, formatting loss modes, draft escape hatch, and connector requirement. An agent has everything needed to call it safely.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents all 11 parameters thoroughly (including hook, brand, recipe, headings enum semantics). The description reiterates the title/body contract and allowLossy/publish behavior, adding framing but little beyond the schema. Baseline 3 is correct when the schema does the heavy lifting.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb (publish) and resource (long-form Article on X), and explicitly distinguishes it from the sibling post_to_x by clarifying it is X's Article format, 'a different thing from a long POST'. An agent can route correctly without inspecting either schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides explicit when-to-use guidance ('show the user the full text and get an explicit yes first'), when to use publish:false (review drafts), and warns against iterating by republishing. The user-approval precondition is spelled out, not implied.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools