Your methods, kept as skills your AI follows, for every client.

A skill is a method you write down once, like your site audit or your content brief, and your AI follows it step by step. Unilytics keeps your skills in one library, with each client's context beside them and your team sharing them. It works in Claude today.

New accounts start with 50 credits. Connecting, and reading your skills, projects and client context, costs nothing; research tools and some writes use credits.

Illustration of a skill in Unilytics: the aeo-audit skill in the Acme Bikes project, with its details, its files (SKILL.md, a reference folder and a data folder) and the first steps of SKILL.md. Beside it, an AI client saves version 13 of the skill and adds a note to the skill's memory.

What a skill is

  • Your steps, in plain words

    A skill holds the steps your AI follows for one job and the thinking behind them. You write it the way you'd brief a colleague, or describe how you work and ask your AI to save it as a skill.

  • Reference files it checks against

    Checklists, scoring tables and other reference files sit with the skill, and your AI checks its work against them as it goes.

  • Notes from past runs

    What a run teaches is saved as notes kept with the skill (its memory), and your AI reads those notes at the start of the next run.

How it works, for every client you serve

  1. 01

    Every client or brand, its own project

    Each client or brand gets a project with its Brand Kit, knowledge base and collaborators, so a skill you run for that client uses its context.

  2. 02

    Each skill is your method, written down

    The steps, the reasoning behind them and the reference files, kept once and used for every client.

  3. 03

    Every save is a version

    Each save becomes a new version with a note on what changed, and a check stops two edits from overwriting each other.

  4. 04

    Your AI works the skill's steps

    Describe the task. Your AI finds the skill that fits, loads the project's context and works through the steps.

  5. 05

    The deliverable the skill promises

    The run ends in the output the skill describes, such as a slide deck or a Word document, ready for the client.

  6. 06

    The skill gets better

    What the run taught is saved with the skill for the next one. When you find a better method, saving it makes the next version.

Where your skills live

Your skills sit in Unilytics, not inside one AI app. Your AI reads them through the Unilytics connector, and you can browse them in the Skill Library in the Unilytics app.

  • My skills

    Skills you or your AI saved. They're private to you, and you can change them.
  • Project skills

    Skills saved to a client's project, which everyone invited to that project can use. Already saved one under My skills? Ask your AI to copy it into the project: yours stays where it is.
  • Ready-made skills

    Skills Unilytics writes and maintains, and a built-in library grouped by topic. Both are read-only: ask your AI to make a copy based on one, and the copy is yours to change.
  • The right skill for the task

    Describe the task and Unilytics returns the best-matching skills, each with the reason it matched, from your own skills, your projects' skills and the ready-made ones.
  • Every version kept

    The Changes tab on your own and your projects' skills shows what changed between any two versions and who saved each one.
  • Yours to take with you

    Click Export on any of your own or your projects' skills to download it as a zip with all its files. Ready-made skills, and copies made from them, can't be downloaded.

Shared with your team, client by client

  • A role for each person

    Invite a colleague to a client's project as a Viewer (uses its skills, changes nothing), an Editor (also changes them through their AI) or an Admin (also archives them and invites people).

  • Who changed what

    Every version of a project skill names the person who saved it, so the team can see how its method changed and why.

  • A link to any project skill

    Open a skill from a project's Skills page and send its link to a collaborator. It opens inside that project for anyone with access.

To invite someone (you need to own the project or be its Admin): open it in the Unilytics app, click Share project, enter their email, pick the role and click Invite. Unilytics doesn't email the invite: click Copy link and send it yourself. Credits aren't pooled: each person's AI uses their own account's credits.

Works with the tools you already connected

You add Unilytics to Claude next to the other connectors you use. When a step of a ready-made skill needs your own data, it tells your AI to use your own connector for that tool, and your AI calls it in the same chat.

  • Your Semrush or Ahrefs account

    A step that needs your Semrush or Ahrefs data uses your connector for that tool, so the data comes from your own plan with it. Without a connector, the skill uses the tool's web app you're signed in to, or an export you share.
  • Search Console through your own access

    Steps that need Search Console data use a Search Console connector you add to your AI, the Search Console website, or an export. Unilytics doesn't get access to your Search Console.

Coming next

Not on every account yet. Everything above this section works today. Available soon means it runs on our own skill library first; Planned means it isn't built yet.

  • Skill Checker

    An AI agent that checks each skill is clear and complete, so any AI runs it the same way, and suggests fixes and the gaps only you can fill in.
    Available soon
  • Librarian

    An AI agent that finds duplicates and old copies in your library and suggests combining or retiring them.
    Available soon
  • Researcher

    An AI agent that reads trusted sources and suggests deeper steps for a skill, quoting each source.
    Planned
  • Agent Inbox

    The agents' suggestions wait in one inbox in the Unilytics app, across your clients. No suggestion changes a skill until a person approves it there: your AI can read and dismiss suggestions, but not approve them. You can also send a draft back or undo a merge.
    Available soon
  • Your own tool servers in Unilytics

    Add a custom MCP server (a server of tools your AI can call) to Unilytics once, and use it in your client projects. Until then, add it to your AI next to Unilytics.
    Planned
  • Tests on every skill

    Example tasks saved with each skill, with what a good result looks like, checked before a new version goes live.
    Planned

Keep your methods in one place. Run them for every client.

Create your account and start on your own today. Plans aren't sold online: for a plan or more credits, including team plans, write to sales@unilytics.ai. Here for SEO and AEO work? The home page shows the ready-made skills and research tools for it.