The personal brand OS for LinkedIn and Twitter: check, improve, schedule and publish posts.
Scored across 15 tools
The tools mostly have clearly distinct purposes: publishing vs scheduling, voice profile vs voice insights, and ideation vs hook generation are well separated by their descriptions. There is some potential overlap between get_analytics/list_posts and between get_voice_profile/get_voice_insights/score_voice_match, but the descriptions clarify the boundaries enough for correct selection.
All tools use snake_case and follow a predictable verb_noun pattern: get_*, list_*, improve_*, publish_*, schedule_*, score_*, search_*, suggest_*. There are no mixed conventions or vague standalone verbs.
Fifteen tools is at the upper end of a typical well-scoped set, but each one maps to a distinct part of the content workflow: analytics, voice analysis, ideation, drafting, scoring, inspiration, publishing, and scheduling. No tool appears redundant or trivial.
The tool surface covers the core lifecycle from inspiration and ideation through drafting, voice checking, improvement, publishing/scheduling, and analytics. Minor gaps remain around managing scheduled posts, such as viewing the queue, cancelling, or rescheduling, and there is no delete/update operation for existing posts.
Get engagement totals (impressions, likes, replies, retweets, post count) and the top post for a time window. Use when the user asks how their posts performed, how yesterday went, or how they're trending.
Inputs: window, platform.
The signed-in VoiceMoat account: its email, its plan, and whether that plan includes connecting to an assistant. Use when the user asks which VoiceMoat account or plan they are connected with.
Inputs: no named parameters.
Fetch full text + engagement numbers for one specific post by id. Use when the user references a specific post and you want exact numbers.
Inputs: post_id (required).
Generate N distinct post ideas on a topic. Use when the user asks for ideas, angles, or different takes on the same subject. Each idea is a short seed, not a full draft.
Inputs: topic (required), count, platform.
The user's latest Voice Lab analysis for one platform (dominant voice patterns, hook styles, sentence rhythm, vocabulary) and their latest drift reading (how recent posts compare with their usual voice). Use when the user asks what their voice patterns are or whether they are drifting. Returns empty results when the user has not run Voice Lab yet; Voice Lab in VoiceMoat is where they run it.
Inputs: platform.
The user's trained voice profile for one platform: writing persona, tone markers, personality traits and phrases they never use. Read-only. Use when the user asks what their voice is like or what to keep and avoid in their writing.
Inputs: platform.
Improve a draft with VoiceMoat's improve engine, the same one behind the Improve panel in VoiceMoat's Content Studio. Two modes: 'full' does one deep pass (grammar, then the opening line, then line breaks, then platform formatting) and returns the improved post with what changed and why; 'quick' returns 2 to 3 alternative rewrites to choose between. Use when the user asks to improve, tighten or fix a specific draft. Costs credits (quick 3, full 5), so it is for text the user asked to improve, once per text. Nothing is saved or posted: the result comes back to the user to accept or ignore.
Inputs: mode (required), content (required), platform, is_thread.
The creators the user has analyzed or learned from on the current platform, with each creator's topics and signature opening-line patterns. Use when the user asks for ideas or a post like the creators they follow or study, or asks who they have been learning from. Works on Twitter and LinkedIn, scoped to the active platform. Returns an empty list when they have not analyzed any creators yet; the Inspiration > Creators tab in VoiceMoat is where they add some.
Inputs: limit.
List the user's best-performing posts (sorted by likes) in a time window. Use when the user asks what worked, what went viral, or for examples of their best content.
Inputs: limit, window.
The Twitter and LinkedIn profiles connected to this VoiceMoat account, each with its account id, handle and display name, plus the platform the tools use when the user does not name one. Use when the user asks which profiles are connected or which one they are on. Each account id is what publish_post and schedule_post take as account_id.
Inputs: no named parameters.
Publish a post to the user's LinkedIn or Twitter account immediately. Call it once WITHOUT confirmation_token to get a preview and a token, show the user exactly what would be posted and to which account, then call it again with the token only if they agree. On Twitter, separating blocks of text with a line containing only three dashes posts them as a chained thread.
Inputs: content (required), platform (required), format, signature, +4 more. Returns: url, kind, parts, action, +9 more
Add a post to the user's scheduled queue, to publish unattended at a time they choose. Call it once WITHOUT confirmation_token to get a preview and a token, then again with the token only if the user agrees. The time must be UTC ISO-8601 and at least ten minutes in the future.
Inputs: content (required), platform (required), scheduled_for (required), format, +7 more. Returns: url, kind, parts, action, +9 more
Check whether a piece of text sounds like the user, as two separate results. Style: how many of their measurable writing habits it matches, as N of M checks (sentence length, emoji, casing, post length, opener, AI tells, drift), with a real line of theirs where one fits. Stance: whether what it says fits the views, stories and framings to avoid in their claim bank, as On voice, Check one line or Off voice, with each flagged line and the claim it hit. There is no blended voice score. Use when the user asks whether a draft sounds like them, is on voice, or is something they would write, including a draft written for them in this conversation. Stance costs 1 credit and runs only when their claim bank has items and the text states something to check; Style is free. For text the user asked about, checked once.
Inputs: content (required). Returns: kind, style, stance, status, +2 more
Search the shared LinkedIn inspiration corpus: public posts from the creators this community tracks, the same set the Inspiration > Explore tab shows. Use when the user asks what is working in their niche, wants examples of posts on a subject, asks how other people open a post about something, or wants to see what a named creator has been posting. LinkedIn only, and free to run. Returns other people's posts, which are reference material and never the user's own draft. An empty result means nothing matched, not that the user has no posts.
Inputs: limit, query, creator, media_type.
Generate 3 scroll-stopping opening lines for a topic, from genuinely different angles (contrarian, story, listicle, reframe, observation). Use when the user asks for a hook, a first line, or doesn't know how to start.
Inputs: topic (required), platform.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Schedule and publish to LinkedIn, X, and Threads from your AI. Content calendar, approval-first.
Editorial tools for LinkedIn. Capture ideas, draft posts, build narrative arcs, manage a schedule.
Draft LinkedIn posts in your voice, then schedule or publish them through LinkedIn's official API.
Personal branding app for LinkedIn, Instagram and YouTube, connected to your own AI.
Add one secure layer between your agents and this server.
