Skip to content

Frequently Asked Questions (FAQ)

Find answers to common questions about using and building with Seer Marketing Agents.

General

What is Seer Marketing Agents?

Seer Marketing Agents is an AI-powered infrastructure designed for SEO, Analytics, and Paid Media workflows. It provides a set of standardized "agents" that automate complex marketing tasks—like content audits and SERP analysis—while maintaining the high-quality standards of the Seer Interactive team.

Do I need paid AI subscriptions?
  • OpenCode (Primary): Uses your configured AI providers. You will need API keys and an active billing account with providers like OpenAI, Anthropic, or Google (via Vercel AI SDK).
  • Claude Code (Secondary): Requires an active Anthropic subscription that includes Claude Code access.
  • MCP Servers: Some data sources (like DataForSEO or BigQuery) have their own separate costs and subscription requirements.
How is my data handled?

Data privacy is a core principle of the system: * Local Processing: Most data processing happens in your local environment. * Direct Connections: MCP servers connect directly from your machine to your data sources (BigQuery, Wrike, etc.). * No Training: By default, data sent to Enterprise AI APIs (like those used by OpenCode) is not used to train external models. Always check your specific provider's terms to confirm.

Using the System

What if an MCP server is unavailable?

The system is designed with graceful degradation. If a required MCP server (like BigQuery) is missing or disconnected, the command will not fail completely. Instead, it will: 1. Notify you that the server is unavailable. 2. Prompt you to provide the necessary data manually (e.g., by pasting CSV content or providing a URL). 3. Continue the workflow using the provided data.

What's the difference between commands and skills?
  • Commands: These are user-triggered workflows. You explicitly run them by typing / followed by the command name (e.g., /content-audit).
  • Skills: These are auto-activated behaviors. They define how the agent thinks and writes (e.g., writing-standards). You don't trigger them manually; the system loads them automatically based on the context of your request.

Commands use skills automatically to ensure the output meets our quality standards.

How long do workflows take?

Workflows are designed to save you hours of manual work, though they do take some time to process: * Content Audit: 30-60 minutes (typically saves 3+ hours) * Search Landscape: 90-120 minutes (typically saves 2+ days) * Content Outline: 45-90 minutes

Technical

Can I use this without MCP servers?

Yes. While MCP servers provide the most seamless experience by automatically fetching data, all commands support manual input fallbacks. You can find more details on connecting data in the MCP Integration Guide.

How do I add new commands or skills?

The system uses a modular plugin architecture. New commands and skills are added by creating or modifying division plugins. * Commands live in plugins/divisions/{division}/commands/ * Skills live in plugins/divisions/{division}/skills/

For a step-by-step walkthrough, see the Plugin Authoring Guide.

What platforms are supported?
  • OpenCode: Our primary platform, offering the most flexibility and multi-provider support via the Vercel AI SDK.
  • Claude Code: Fully compatible via the Agent Skills Spec, allowing you to use these agents directly within the Claude CLI.

Troubleshooting

Why isn't my command working?

If a command isn't responding as expected, check the following: 1. Directory: Ensure you are running the agent from the root of the agents-infra directory. 2. Plugin Loading: Run /utils:commands to see if your plugin is active. 3. Core Dependencies: Verify that the core-dependencies plugin is present and updated.

For more detailed solutions, visit the Troubleshooting Guide.