# Changelog of AI SEO & GEO MCP Server: SERP API and keywords (`mrbridge/seo-geo-mcp`) Actor

- **URL**: https://apify.com/mrbridge/seo-geo-mcp/changelog.md
- **Full Actor documentation**: https://apify.com/mrbridge/seo-geo-mcp.md

## Changelog

All notable public changes to SEO & GEO MCP are documented here.

#### v0.2.6 (2026-08-13)

🐛 Bug Fixes

- Restored the optional Target, Location, Language, and Example Prompt fields required to render published Example pages.
- Clarified that these fields are Example-page metadata and do not configure the Standby MCP server or its tool calls.
- Removed the remaining Store title from the Actor manifest so Console publication fields remain the source of truth across builds.

#### v0.2.5 (2026-08-12)

🐛 Bug Fixes

- Added a lightweight `GET /mcp` event stream for compatibility with MCP clients that probe or listen on the endpoint before sending JSON-RPC requests.
- Kept Streamable HTTP `POST /mcp` as the request transport and excluded the compatibility stream from Tool Read billing.
- Published the provider-neutral SEO & GEO MCP server identity during MCP initialization.

#### v0.2.4 (2026-08-12)

🐛 Bug Fixes

- Added the standards-compliant `Allow: POST` header to unsupported `GET /mcp` responses for stricter MCP client compatibility.
- Clarified that opening the MCP endpoint like a web page exercises HTTP GET and is not a complete MCP connection test.

#### v0.2.3 (2026-08-12)

🐛 Bug Fixes

- Simplified Actor Input to match the connection-first setup used by Amazon MCP Server.
- Removed the Target, Location, Language, and Example Prompt worksheet fields because MCP requests receive those values directly from the connected AI assistant.
- Kept the exact Streamable HTTP endpoint, Apify token instructions, available-data summary, OAuth guidance, and cold-start notice directly above the empty form.

#### v0.2.2 (2026-08-12)

🐛 Bug Fixes

- Clarified the billing unit as **Tool Read**: a tool call is a request, while each non-empty scalar data point returned by that call is one Tool Read.
- Kept the Free-plan base rate at **$0.005 per Tool Read ($5 per 1,000 data points read)**.
- Added four provider-neutral Input worksheet fields—Target, Location, Language, and Example Prompt—so Actor and published Task pages have a useful Input block without restoring credentials or module selection.
- Explained that worksheet values must be copied into the connected AI assistant and are not executed by clicking Start.

#### v0.2.1 (2026-08-12)

⚠️ BREAKING CHANGE

- Replaced per-successful-tool-call billing with one **Data Point** event for every non-empty scalar value in the useful SEO or GEO result payload.
- Set the Free-plan base rate to **$0.005 per data point ($5 per 1,000)** while retaining the automatic plan discounts shown in the Pricing tab.
- Excluded response metadata, JSON containers, `null` or empty values, failed calls, empty results, and MCP protocol traffic from data-point billing.
- Added exact batch-count charging, empty-result protection, spending-limit handling, and offline tests for cross-module response shapes.

#### v0.2.0 (2026-08-11)

⚠️ BREAKING CHANGE

- Renamed the Actor and public product from its provider-specific identity to **SEO & GEO MCP**.
- Changed the public Standby endpoint to `https://mrbridge--seo-geo-mcp.apify.actor/mcp`.
- Simplified client authentication: users now supply only their Apify API token, either in `?token=YOUR_APIFY_API_TOKEN` or as a Bearer header.
- Removed provider login and password fields from public Actor Input. The Actor owner now manages the upstream service configuration.
- Prevented former credential headers and query parameters from overriding the owner-managed account.
- Added provider-neutral public tool-group labels, tool aliases, copy, schemas, connection instructions, and error messages.
- Removed module-selection fields and URL filtering; the complete tool catalog is now always exposed.
- Preserved `tool-call` as the primary Pay-per-Event event, titled **Tool Call**, charged once only after a confirmed successful tool result.
- Added plan-based pricing details and direct setup instructions for Claude, ChatGPT, Cursor, VS Code, Claude Code, and other Streamable HTTP clients.

#### v0.1.1 (2026-08-11)

📝 Documentation

- Added direct and Task-specific connection guidance to Actor Input.
- Added the complete MCP URL with an uppercase Apify token placeholder.
- Expanded Input connection guidance and Output-tab links.

#### v0.1.0 (2026-08-07)

🚀 Initial release

- Added an Apify Standby Streamable HTTP MCP endpoint at `/mcp`.
- Added ten tool modules for search results, keywords, SEO research, backlinks, on-page audits, local business, domain intelligence, content, e-commerce, and AI visibility.
- Added structured errors, safe logging, schemas, mock-only tests, and Pay-per-Event charging after confirmed successful calls.