> For the complete documentation index, see [llms.txt](https://docs.nested-knowledge.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.nested-knowledge.com/get-started/ai-documentation/smart-instructions.md). # Smart Instructions Learn how to use smart instructions to set up your nest. **Smart Instructions** is where you tell the AI tools in your nest what they need to know beyond your criteria and tags. It contains your **Research Question** and four sets of instructions, one each for screening, extraction, critical appraisal, and synthesis. Every AI tool at that stage reads them. Without it, guidance that applies across your whole review has to be repeated: the same definition of "adult" in several screening criteria, the same time points in 10 outcome tags, or a long research question carrying details it was never meant to hold. **Smart Instructions** lets you state that guidance once, keep your research question to one or two sentences, and change the guidance in one place. Use it for guidance that applies across a whole stage. Keep what decides eligibility in your screening criteria, and what defines a single data point in that tag's question. **Smart Instructions** adds context to criteria and tags; it doesn't replace them. {% hint style="info" %} Instructions determine what your AI tools produce. Record them in your methods alongside your research question. {% endhint %}

The Smart Instructions page, with fields for your research question and stage-specific instructions. — Use Smart Instructions to add guidance for your research question, screening, extraction, critical appraisal, and synthesis.

### Set your Smart Instructions {% stepper %} {% step %} ### Open Smart Instructions On **Nest Home**, select **Smart Instructions**, beside **Nest Protocol**. You can also select **Edit** in any AI tool's **Smart Instructions** panel, which opens this page in a new tab.

{% endstep %} {% step %} ### Write Or Refine Your Research Question Enter your research question in **Research Question**. To have AI suggest a tighter wording, select **Refine**, answer any follow-up questions that apply, and review the suggestion before saving. Keep it to one or two sentences. Move definitions, formatting preferences, and other detail into the instruction fields below. {% endstep %} {% step %} ### Add Instructions For Each Stage Fill in only the fields you need. Each field is read by the AI tools at that stage: * Screening Instructions: **Smart Screener, Smart Config: Screening** * Extraction Instructions: **Core Smart Tags, Smart Config: Tagging, Adaptive Smart Tags, Smart Meta-Analytical Extraction** * Critical Appraisal Instructions: **Smart Critical Appraisal** * Synthesis Instructions: **Smart Insights, Smart Abstract** Changes save automatically. **Smart Search** uses your research question but not these instructions. One field covers both Abstract and Full Text Screening. The AI knows which it's working on, so you don't need separate abstract and full-text instructions {% endstep %} {% step %} ### Check Them Before You Run A Tool When you open an AI tool, its modal shows the **Research Question** and the stage's instructions it will use. Check them before you select **Generate**. {% endstep %} {% endstepper %} ### Decide What To Put In Smart Instructions Put guidance here when it applies across a whole stage and you'd otherwise repeat it. For example: * **Shared definitions.** "Adults means participants aged 18 or over." "Treat GLP-1 receptor agonists as one intervention class." * **Scope that applies to every tag or criterion.** "Extract outcomes only at 12 and 52 weeks." * **Format, tone, and language.** "Write rationale in Spanish." "Use past tense and third person." Leave it in your criteria or tags when it applies to one question only. A criterion still decides eligibility; a tag question still defines one data point. Be specific. More detail generally helps more than it hurts, and the AI decides which instructions apply to each criterion or tag. You can write broad rules, such as "for any outcome, extract the time point too," without naming every outcome. ### Understand How Instructions Affect Output **When instructions conflict.** If a criterion or tag says something different from your **Smart Instructions**, the criterion or tag generally takes precedence, because it's more specific to the task the AI is addressing. The handling of conflicts is inherently inconsistent. **Rationale versus annotations.** Rationale text follows your instructions. If you ask for Spanish, **Smart Screener** writes its rationale in Spanish. Annotations always highlight the original text in the abstract or full text. Rationale can also refer to your instructions when explaining an answer. **Changing instructions later.** Editing Smart Instructions does not trigger automatic reprocessing. You will need to rerun the corresponding AI tool so answers will reflect updated instructions. ### FAQs

Does Smart Instructions replace my screening criteria?

No. Your criteria still decide eligibility. **Screening Instructions** add context that applies across all criteria, such as shared definitions or the language of rationale text.

Do I need separate instructions for abstract and full-text screening?

No. One **Screening Instructions** field covers both. The AI knows which it's screening.

What happens if my instructions contradict a tag or criterion?

The tag or criterion generally takes precedence, since it's more specific. Remove the conflict rather than relying on that.

Does Smart Search use my instructions?

No. **Smart Search** uses only your research question.

If I change my instructions, are existing AI answers updated?

No. AI tools will need to be rerun after instructions are changed.

### Next Steps


Smart Search	Generate and review PubMed search queries from your research question.	Smart Search
Smart Screener	Generate and review screening answers using your screening instructions.	Smart Screener
Adaptive Smart Tags	Generate and review extracted data using your extraction instructions.	Adaptive Smart Tags
Smart Critical Appraisal	Generate and review appraisal recommendations using your critical appraisal instructions.	Smart Critical Appraisal
Smart Insights	Generate, review, and export insights using your synthesis instructions.	Smart Insights

--- # 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.nested-knowledge.com/get-started/ai-documentation/smart-instructions.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.