SuperAPI AI Directory
Partner with us中文AI newsFavorites
All toolsOther AI toolstoprank

toprank

toprank (repository name nowork-studio/notfair-plugin) is an open-source collection of SEO, GEO, and marketing skills for AI agents, licensed under MIT. It is not a generic prompt collection, but rather provides executable, reviewable marketing workflows for compatible agents such as Claude Code, Codex, Hermes, Cursor, Kiro, and Gemini CLI. The repository currently contains 48 skills, covering SEO, GEO, paid media, advertising platforms, analytics, WordPress, CRM, and cross-model review, among other areas. Each skill has a unified skills/<name>/SKILL.md entry point, and describes its workflow through a WORKFLOW.md in a category folder, along with reference documents, scripts, and evaluation cases. Users can have agents audit websites, investigate the causes of traffic declines, analyze GA4 and Search Console data, troubleshoot wasted ad spend on platforms such as Google, Meta, X, LinkedIn, Reddit, and TikTok, operate connected WordPress and GoHighLevel accounts, create campaign plans, and produce reviewable changes. The repository is compatible with the Agent Plugins 1.0 specification and accesses real-time platform data through a universal MCP connection called NotFair. Both connections and account operations require the user to complete browser OAuth authorization.

Open website

About this tool

Purpose and Positioning

toprank (repository identifier nowork-studio/notfair-plugin) is an open-source collection of SEO, GEO, and marketing skills for AI agents, licensed under MIT. Its goal is not to provide another generic prompt collection, but to give compatible agents such as Claude Code, Codex, Hermes, Cursor, Kiro, and Gemini CLI marketing workflows they can actually follow.

Main Capabilities

The repository currently contains 48 skills, spanning categories such as SEO, GEO, paid media, advertising platforms, analytics, WordPress, CRM, and cross-model review. With these skills, agents can:

  • Audit websites, or investigate the causes of organic traffic declines;
  • Analyze GA4 and Search Console data;
  • Troubleshoot wasted ad spend on platforms such as Google, Meta, X, LinkedIn, Reddit, and TikTok;
  • Operate connected WordPress and GoHighLevel accounts;
  • Create marketing campaign plans and produce reviewable changes;
  • Find pages likely to be cited in AI answers (GEO-related).

Each skill has a unique discoverable entry point, skills/<name>/SKILL.md, which loads the readable WORKFLOW.md in the corresponding category folder and optionally includes reference materials, scripts, and evaluation cases. Users can inspect, adjust, or contribute better workflows.

How to Use

The repository is compatible with the Agent Plugins 1.0 specification, and the plugin.json and mcp.json in the root directory follow that specification. Different installation paths can be used depending on the host environment:

  • Install the NotFair plugin from the marketplace, then directly request the desired workflow; natural language can also be used, for example, "audit my website and explain why organic traffic is declining," "find pages likely to be cited in AI answers," or "check last month's ad spend and point out the safest optimization opportunities for improving ROAS."
  • In Cursor, install from the plugin marketplace; the plugin loads this repository's open-source marketing skills and registers a hosted NotFair MCP connection. After completing the browser OAuth flow as prompted by Cursor, it is ready to use. Before the marketplace listing is available, you can clone the repository to Cursor's local pluginCatalog for local testing; after installing or updating a local plugin, Cursor must be restarted.
  • In Kiro, install the public repository as a custom power from the Powers panel; once the listing is approved, NotFair can also be found in the Powers registry.
  • In Gemini CLI, the extension can be installed directly from GitHub; the extension loads the repository guidance and the general NotFair MCP, and when authentication is required, run /mcp auth NotFair.
  • In Codex, the general NotFair plugin can be installed directly through Codex: Codex will install the skills, register a NotFair MCP connection, and open its OAuth flow. If you prefer a workspace-local checkout, you can clone the repository and open it as a workspace, and AGENTS.md will map marketing requests to the appropriate skills. If the nowork-studio marketplace is already configured, you can instead refresh that marketplace.

For host-specific configuration, you can provide INSTALL_FOR_AGENTS.md to the agent, or paste the corresponding content according to the instructions in that file.

Design Approach

The repository chooses "skill splitting" rather than a "single giant marketing agent" because marketing tasks become unreliable under the same set of vague prompts. The NotFair plugin breaks the work into focused, testable workflows:

  • Specialization: each skill has a clear responsibility, required inputs, decision rules, and output format.
  • Evidence-oriented: real-time data workflows use Search Console, Google Analytics, Google Ads, Meta Ads, X Ads, LinkedIn Ads, Reddit Ads, TikTok Ads, WordPress, or GoHighLevel, rather than guessing based only on general best practices.
  • Safe by default: read-only review precedes changes, paid media changes remain explicit, and unsupported capabilities are not implied.
  • Host-agnostic: canonical skills are plain files, not logic locked into a particular agent runtime.
  • Reproducible: all are MIT licensed, and teams can review and adjust the workflows to meet their own standards.

It also includes content related to cross-model review and maintenance; for example, Gemini review, after completing authentication setup, sends the selected review context to Google's Gemini CLI.

Real-Time Data and Integrations

Some skills are based entirely on the repository, URL or a provided export file; live account analysis uses a generic NotFair MCP connected via OAuth. The plugin registers only one server named NotFair, shared across all supported platforms.

The agent selects tools based on instructions and capability descriptions from the live server. Skills only provide marketing workflows and evidence standards, and do not hardcode MCP tool names or call order. For manual configuration and migration from the old per-platform connections, refer to the connection and upgrade guide. Each platform must be connected in the NotFair workspace selected during OAuth before it becomes available, and the account setup guidance returned by the server should be followed.

Supported account operations depend on the live connection. Changes must remain within the scope of user authorization and be verified against the account's actual state. Unless the current agent session exposes a verified connector, Amazon and ChatGPT Ads are used only for planning or export review workflows. setup-cms is a credential wizard for local SEO scripts; live WordPress work uses /notfair:wordpress.

Within skills, connectors use tool-agnostic placeholders, such as ~~google-ads, ~~meta-ads, ~~x-ads, ~~linkedin-ads, ~~reddit-ads, ~~tiktok-ads, ~~search-console, ~~google-analytics, ~~wordpress and ~~gohighlevel. The agent resolves each placeholder to a compatible tool available in the current session, so workflows are not tied to a single MCP namespace.

Privacy and Local Workflows

The plugin connects via browser OAuth to the declared NotFair MCP endpoint. Requested tool parameters are sent to NotFair, and results from connected providers are returned to the authorized AI client. For storage, retention, and deletion details, see the privacy policy. It is recommended to connect only the workspaces and platform accounts you plan to use.

Optional local workflows run readable Python or shell scripts. The crawler requests user-provided URLs; Search Console direct scripts use Google Cloud login explicitly authorized by the user; CMS scripts use site credentials configured via setup-cms; Gemini review, after completing authentication setup, sends the selected review context to Google's Gemini CLI. These workflows require explicit selection, credentials should not appear in reports or commits, and live platform work should preferentially use an available OAuth MCP. The plugin has no automatic hooks or local MCP server. The calendar viewer serves local planning data on the loopback address, and the optional update checker reads public release files from GitHub.

The formerly standalone NotFair Next.js blog repository has moved to its own public repository and is distributed via GitHub Releases; its source code and release process are no longer packaged with this AI agent plugin. The deprecated NotFair CMO app is also no longer included.

Repository organization and contribution

AGENTS.md is the general entry point, responsible for mapping user intent to the canonical SKILL.md and explaining the external dependencies required by each workflow. Each skill is located in its own category folder.

When adding or modifying a skill: keep the workflow focused on one clear marketing task; use imperative, testable instructions in WORKFLOW.md and keep the entry points under skills/ in sync; add or update evaluation coverage; update AGENTS.md, the plugin manifest, VERSION, and CHANGELOG.md; submit pull requests as "one skill or one coherent improvement." Repository conventions are detailed in the plugin development documentation.

Applicable scenarios

Suitable for teams that want AI agents to execute SEO, GEO, paid media, and analytics marketing tasks according to reviewable processes, especially in scenarios where workflows need to be reused, forked, or adjusted to their own standards.

Limitations and information not yet provided

The current materials do not provide specific pricing, billing methods, or paid plan information; they do not explain specific input and output examples for each skill; they do not list the complete inventory of all 48 skills; the specific account operation details supported by live features depend on the runtime connection and are not enumerated one by one in the materials. Amazon and ChatGPT Ads are limited to planning or export review when a verified connector is missing. Before use, you must confirm the scope of authorization and credential management requirements yourself.

Related tools

Browse category