Product updates

Changelog

Dated notes on Cralo releases, product improvements, and operational changes.

Release notes

  1. Remote MCP is public in the docs

    You can add Cralo as a remote MCP server at https://mcp.cralo.ai/mcp in Cursor, Claude, ChatGPT, or Grok.

    • The docs include a Connect with MCP guide with per-host setup for remote Streamable HTTP.
    • The overview, quickstart, and local MCP pages point at the remote connector first.
    [Connect with MCP]

  2. Cralo is live at cralo.ai

    Cralo is the new product name and cralo.ai is the new home for the marketing site, documentation, and agent setup.

    • The website, documentation, app navigation, legal pages, emails, and public product metadata now use the Cralo name.
    • The @opentooler/cli package, opentooler command, and existing app and API endpoints remain compatible throughout the transition.
    [Visit Cralo][Read the docs]

  3. Private product analytics and unified error tracking

    The marketing site, app, and API now report the product journey to PostHog, with errors handled there instead of Sentry.

    • New events cover acquisition, onboarding, catalog discovery, tool use, billing, and API outcomes without sending prompts, payloads, tokens, emails, or full URLs.
    • Browser and backend errors are redacted before capture, source IP storage is disabled, and Sentry has been removed.

  4. Apify pass-through pricing and catalog reprices

    Apify scrapers now quote from the requested limit up to 1000 results and settle from returned rows, and TikHub, SerpApi, and DataForSEO prices match current provider cost.

    • Instagram posts and Apify tools use per-result Starter PPE pricing, with long-running scrapers on the queued worker.
    • Catalog summaries now say when price is per result, per URL, or per minute.

  5. Search tool output schema hotfix

    Brave search, Google search, Google Jobs, YouTube search, and TikTok posts no longer fail after a successful provider call.

    • Published output schemas again include the description field those adapters return.
    • Google search now unescapes malformed SerpApi result URLs so they validate.

  6. Catalog pricing, limits, and Brave search

    Live tools now match provider pricing and request limits, Hunter and ScrapeCreators are retired, and Brave web search is in the catalog.

    • Firecrawl, Exa, Perplexity, SerpApi, and scraper tools were resynced to current provider capabilities.
    • Hunter email finder and ScrapeCreators TikTok profile are no longer available.

  7. Centered logo mark

    The square logo now sits on the geometric center of the canvas, instead of the unused descender space in the font em box.

    • Favicons, the in-app logo mark, and the brand lockup share the same vertically centered mark.

  8. Clearer homepage, catalog search, and public legal pages

    A first visit now explains what Cralo is, how to start, and how billing works, then points to privacy, terms, and support.

    • The homepage states the product in plain language, with how-it-works and use-case sections.
    • The tools catalog is searchable, and privacy, terms, and support are public site pages.

  9. Cralo CLI 0.2.0

    The published CLI now includes an MCP server, so Claude Code, Claude Desktop, and Cursor can use Cralo without shelling out for every call.

    • MCP serve exposes catalog search, describe, call, invocation status, and balance over stdio.
    • A new integrations configure command prints or writes host setup for Claude Code, Claude Desktop, and Cursor.
    [View on npm]

  10. Cralo CLI 0.1.7

    CLI output became easier to scan while keeping the complete result available in JSON.

    • Search and people results now use concise, readable terminal summaries.
    • Production deployment validation checks that active fast and queued tools have their required adapters.
    [View on npm]

  11. Payment history and clearer usage navigation

    The dashboard now separates billing from invocation usage and shows past payments on the billing page.

    • Payment history includes the paid amount, credits granted, status, and payment identifier.
    • Usage remains available from its own dashboard navigation entry.

  12. More reliable YouTube transcript downloads

    The CLI media path added a pinned local token provider and stronger download checks for YouTube transcription.

    • Downloaded helper components are checksum verified before use.
    • Preflight and retry handling now surface clearer media download failures.

  13. Cralo CLI published on npm

    The public Cralo command-line package launched with tool discovery, invocation, authentication, and billing commands.

    • Install the CLI globally from the public @opentooler/cli package.
    • The marketing site and public agent skill now point to the published CLI flow.
    [View package]