> For the complete documentation index, see [llms.txt](https://docs.zerve.ai/zerve-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.zerve.ai/zerve-docs/agents-in-the-workspace/skills.md). # Skills Reusable instructions and best practices the agent follows, built in or written by your team. A **skill** is a reusable set of instructions and best practices the agent can follow, such as how to profile a new dataset or how your team backtests a strategy. Skills help the agent work the way your team works, consistently, without you repeating the same instructions in every conversation. Manage skills in **Settings → Skills**. ## Default and custom skills | Type | Details | | ------------------ | ------------------------------------------------------------------------------- | | **Default skills** | Built into Zerve, covering common data and research work. They can't be edited. | | **Custom skills** | Written by your team. You can create, edit and delete them. | Both are listed together under **My Skills + Defaults**. ### Default skills Zerve includes these skills: | Area | Skills | | --------------------------------- | ----------------------------------------------------------------------------------------------------------- | | **Data analysis and exploration** | Data profiling and quality; Exploratory data analysis | | **Quantitative research** | Statistical analysis and hypothesis testing; Time series analysis; Backtesting and financial risk metrics | | **Dashboards and reporting** | Report structure and KPIs; Parameterized and repeatable reporting | | **SQL and data warehousing** | SQL best practices; Data discovery and schema exploration; Warehouse performance and large tables | | **ML modeling and tuning** | Feature engineering; Model selection and validation; Hyperparameter tuning; Model evaluation and deployment | ## How the agent uses skills At the start of each piece of work, the agent sees a list of available skills, with each skill's name and description. It then loads the full instructions for any skill it needs. So a good **description** matters: it's how the agent decides when a skill applies. Some skills are always loaded in full: * **Pinned skills**, which apply to everything the agent does * **Skills you attach** to a message, using **Use Skills** in the composer's **+** menu Superagents' planners also see your skills when they draft a plan. ## Creating a skill Select **Add Skill**. A skill is written in Markdown, with a short header giving its `name` and `description`, followed by the instructions themselves. This format is known as SKILL.md. ```markdown --- name: factor-backtest-standards description: How we backtest equity factors. Use whenever building or evaluating a factor or signal. --- # Factor backtest standards - Use point-in-time data only. Never use restated fundamentals. - Lag all signals by at least one trading day. - Report results net of 10 bps transaction costs. - Always show performance for the full period and for the last three years separately. - Hold back the most recent two years as an out-of-sample test. - Report Sharpe ratio, maximum drawdown, turnover and hit rate. ``` Write the description to say both **what** the skill covers and **when** to use it. ## Sharing skills with your team | Option | What it does | | ------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | | **Add to the agent list** | Teammates can choose the skill from the agent's skill menu. Removing it from the list hides it from the menu but doesn't delete it. | | **Pin** | The skill is always applied to everything the agent does, so nobody needs to attach it. | Pin only the skills that should apply to all work, since every pinned skill is loaded every time. ## The agent can write skills for you When the agent works out a useful way of doing something, it can **draft a new skill** from it. The draft appears in the conversation, and you can save it to your skills from there. Nothing is saved until you choose to. For example, after a long session refining how to clean a vendor's data feed, you could ask: > Turn what we just worked out into a skill, so the team can reuse it. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.zerve.ai/zerve-docs/agents-in-the-workspace/skills.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.