360 Secure SkillHub
360 安全技能中心

PostKing · Brand Voice

该技能通过 PostKing 的 MCP 工具列出已保存的品牌声音档案,将其应用到内容生成或已有文本改写中,并在交付前做一次去套话处理,用更平实具体的表达替换 LLM 陈词滥调与填充语;同时提供 AI 味的诊断性标记。声音档案的创建与导入需在网页端完成,本技能不覆盖。

bitsandtea
bitsandtea
热度 32.69
最近更新 2026-08-26 21:35:22
分类:营销依赖:MCP 服务 / 第三方 API应用场景:文案创作 / 校对润色 / 品牌素材 / 社交媒体

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add postking-brand-voice
Metadata
name
postking-brand-voice
description
List and apply saved PostKing voice profiles when generating or rewriting content, and run a de-slop pass that strips LLM clichés and filler.
license
MIT-0
compatibility
Works with any MCP-compatible client connected to postking-mcp (local stdio or hosted at https://mcp.postking.app/mcp); the pking CLI is an optional fast path when a shell is available.
metadata
{"icon":"https://raw.githubusercontent.com/bitsandtea/postking-skills/main/assets/icons/postking-brand-voice.svg","category":"marketing"}

Brand Voice & Quality

Keep generated and rewritten content on-brand and readable — list saved voice profiles, apply one to a generation or rewrite, and run a de-slop pass that strips LLM clichés and filler before it ships.

Voice-profile creation is web-only. Building a new voice profile (from a writing sample, imported social posts, or a deep voice for a key author) happens in the PostKing dashboard's voice-profiles page — there is no MCP tool to create, extract, or import a voice. This skill covers what comes after: listing an existing profile and applying it.

When to Use

Use this skill when a user wants existing content rewritten in a specific saved brand voice, wants a voiceId to pass into another skill's generation call, or wants a de-slop pass to strip LLM clichés and filler before it ships.

When NOT to Use

Not for creating a new voice profile — that's web-only, done in the PostKing dashboard. Not for generating original content from scratch — that's postking-social/postking, postking-seo, postking-storylines, or postking-reddit, each of which accepts a voiceId from this skill as a parameter. Not for presenting AI-generated content as human-written where a platform, client, publisher, or law expects disclosure.

Minimal tool subset

This skill needs only these 5 tools out of PostKing's 200+:

  • list_voices — list saved voice profiles with their IDs.
  • rewrite_with_voice — rewrite arbitrary text in a specific saved voice.
  • rewrite_text — general rewrite, optionally applying a voice profile.
  • humanize_text — de-slop pass: replaces stock LLM phrasing (filler, hedges, canned transitions) with plainer, more specific prose.
  • check_ai_content — diagnostic: flags passages that read as generic or templated, so a human can revise them.

Prerequisites

  1. Authenticated with credits — see the postking-getting-started skill.
  2. A brand already onboarded — see the postking-getting-started skill.

Voices get applied downstream in the postking skill (posts, blogs, landing pages, repurposing), the postking-seo skill (article writing), and the postking-storylines skill (campaign drafts).

Procedure

"Rewrite this in our brand voice and cut the AI clichés."

  1. Build a voice (dashboard only, one-time). If no voice profile exists yet, tell the user to build one in the PostKing dashboard's voice-profiles page (from a writing sample, imported social posts, or a deep author voice) — there is no MCP tool for this step.
  2. List voices. list_voices({}) — get the profileId for the voice to use.
  3. Apply at generation time (preferred). Pass that voice id as the voice/voiceProfileId param directly on generate_post, generate_blog_post, generate_landing_page, repurpose_content, or reddit_rewrite — no separate rewrite pass needed. These tools live in the postking, postking-seo, postking-storylines, and postking-reddit skills.
  4. Apply to existing text. rewrite_with_voice({ profileId, text, platform }) — or rewrite_text({ text, voice: profileId, platform }) for a lighter-touch pass.
  5. De-slop. humanize_text({ text, platform }) — replaces stock LLM phrasing and clichés with plainer, more specific prose. Run this after voice rewriting, right before the text ships.
  6. Spot-check (optional). check_ai_content({ text }) — flags passages that still read as generic or templated. Treat it as a diagnostic pointing at weak writing for a human to revise, not a number to optimise: do not loop humanize_text against the score.

CLI fast path

pking voice list lists saved voices (equivalent to list_voices). Beyond that, the MCP path above is primary — do not assume additional pking flags exist for rewrite/humanize without checking pking --help first.

Expected Output

Content rewritten in a chosen brand voice, with stock LLM phrasing and filler replaced by plainer, more specific prose.

Disclosure

These tools improve writing quality. They do not change the fact that the content was AI-generated, and must not be used to conceal AI authorship where a platform, client, publisher, or law requires it to be disclosed.

Pitfalls

  • list_voices returns empty — no voice profile exists yet; build one in the PostKing dashboard (creation is web-only, not available via MCP).
  • INSUFFICIENT_CREDITS — surface the checkoutUrl from the error envelope and stop, or hand off to the postking-getting-started skill's billing section (billing_list_packs → billing_topup).
  • RATE_LIMITED — back off using the retryAfter value from the error envelope before retrying.
  • UNAUTHORIZED — re-run the login flow described in the postking-getting-started skill.

Verification

  • list_voices({}) should return successfully; a non-empty result confirms at least one profile exists to reference.
  • After humanize_text, the result should come back successfully and read as plainer and more specific than the original — judge the copy itself, not a detector score.