Accelerate AI Toolkit

The Accelerate AI Toolkit turns Abilities API calls into workflows you run by asking a question. No queries, no dashboards to learn — ask "how did my site do this week?" and get an answer grounded in your real data, with a suggested next action.

Beyond analyzing and testing, the toolkit can now also generate on-brand content and evolve a block over many rounds — composing the same abilities, with every change confirmed unless you explicitly ask it to run on autopilot.

Install#

Install the toolkit as a Claude Code plugin with one command:

claude plugin install accelerate-ai-toolkit

Then connect it to your site:

/accelerate-connect

The connect command walks you through everything — your site URL, your WordPress username, and creating an application password — and stores the credentials securely. Run /accelerate-status any time to check the connection is healthy.

Workflows#

Once connected, just ask. The toolkit routes your question to the right workflow:

The toolkit also includes a built-in abilities reference, so the AI always knows exactly what your site can do.

Sample questions#

• "Which landing page needs the most work?"
• "What should I do first this week?"
• "Set up that A/B test."
• "Did the homepage test actually work?"
• "Make me a bolder hero."
• "Design a new pricing section."
• "Keep improving my homepage hero, round after round."
• "Put the waitlist block on autopilot."

You don't need to know which workflow handles what — phrase the question naturally and the toolkit picks the right one.

Generate and evolve#

The toolkit doesn't just tell you what to test — it can write the variants and run the loop for you.

• Design composes a fresh, on-brand version of a block, or a whole new section, from a brief like "make me a bolder hero." It grounds every variant in your site's entire design grammar — global styles, every synced pattern, representative pages, and your real media — rather than guessing from a single page. It always offers to put the result in an A/B test, but never forces one.
• Evolve runs a multi-round optimization loop on one block. Each round fields bold new challengers, tests them, keeps the winner, and builds the next round on it — until the improvements stop. The number of variants per round scales with your traffic, so every arm still gathers enough data to be measured.

Evolve runs attended by default — you confirm each round's challengers and each winner — or on autopilot, where you grant a standing approval up front and it iterates unattended, reporting progress as it goes. Autopilot removes the per-step click, not the discipline: a winner is still only declared at 95% probability-to-win with enough conversions on every arm, and inconclusive rounds keep running rather than crowning a false positive. Every change is confirmed unless you explicitly ask for autopilot — see Permissions.

As it works, the toolkit keeps a per-site learning journal, so what wins on your site shapes the concepts it proposes next time.

Toolkit resources#

• Repository
• Installation guide
• Authentication guide
• Ability reference
• Skill development guide

For the launch article and broader story, see Closing the loop: the Accelerate AI Toolkit.