> ## Documentation Index
> Fetch the complete documentation index at: https://support.metaview.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Metaview AI Skills

> Install reusable recruiting workflows powered by the Metaview MCP.

Metaview AI Skills give your AI assistant repeatable workflows for preparing interviews, organizing debrief evidence, researching talent markets, launching roles, and supporting sourcing work.

Unlike a one-off prompt, a skill defines a consistent process: what information to gather, which Metaview tools to use, which safeguards to follow, and how to structure the output.

<Info>
  Skills retrieve, organize, and synthesize the information available to you through Metaview. They support human judgment; they do not make hiring decisions or replace your review.
</Info>

## How skills work

<Columns cols={3}>
  <Card title="Ask naturally" icon="message">
    Describe the outcome you need without building a complex prompt.
  </Card>

  <Card title="Follow a workflow" icon="list-check">
    The skill gathers relevant context and follows a defined, reusable process.
  </Card>

  <Card title="Use Metaview context" icon="database">
    The Metaview MCP retrieves the conversations and recruiting data available under your existing permissions.
  </Card>
</Columns>

## Before you begin

You will need:

* A Metaview workspace
* Access to the relevant conversations and recruiting data
* The Metaview MCP connected to your AI assistant — see [MCP Overview](/integrations/mcp-integration/mcp-overview)
* An AI client that supports Skills or reusable project instructions

<Note>
  What a skill can retrieve depends on your Metaview permissions, workspace configuration, connected systems, and the information available in the underlying records.
</Note>

## Install a skill

<Tabs>
  <Tab title="Claude" icon="sparkles">
    <Steps>
      <Step title="Connect Metaview">
        Connect the Metaview MCP using the options available in your Claude workspace.
      </Step>

      <Step title="Copy the skill">
        Open the skill page you want and copy the full `SKILL.md` contents from the **Install this skill** section.
      </Step>

      <Step title="Add the skill">
        Save it as a `SKILL.md` file and add it using the Skills or project configuration available in your Claude workspace.
      </Step>

      <Step title="Ask naturally">
        Request the outcome described by the skill — for example, “Prepare me for my interview with Priya.”
      </Step>
    </Steps>
  </Tab>

  <Tab title="ChatGPT" icon="openai">
    <Steps>
      <Step title="Connect Metaview">
        Add the Metaview MCP connection or app to your ChatGPT workspace. Your workspace administrator may need to enable or approve it.
      </Step>

      <Step title="Copy the skill">
        Open the skill page you want and copy the full `SKILL.md` contents from the **Install this skill** section.
      </Step>

      <Step title="Add the skill">
        Install it through the Skills options available in your workspace.
      </Step>

      <Step title="Use project instructions as an alternative">
        Where native Skills are unavailable, add the workflow to a dedicated Project’s instructions. This may not provide exactly the same behavior as an installed Skill.
      </Step>
    </Steps>
  </Tab>
</Tabs>

<Note>
  The exact installation steps and feature availability depend on your AI client, plan, and workspace settings.
</Note>

## Available skills

### Works with the AI Notetaker

These skills use interviews and conversations already captured in Metaview. Some outputs also depend on scorecards, candidate records, or other connected data being available.

<Columns cols={2}>
  <Card title="Interview Prep" icon="clipboard-list" href="/ai-skills/interview-prep">
    Prepare a factual brief from previous interviews, notes, scorecards, and role context — covering what has already been discussed and what remains unknown.
  </Card>

  <Card title="Interview Debrief" icon="messages" href="/ai-skills/interview-debrief">
    Organize interview evidence by competency, surface agreements and contradictions, and prepare a structured pack for a human-led debrief.
  </Card>

  <Card title="Day-One Brief" icon="sunrise" href="/ai-skills/day-one-brief">
    Turn an accepted candidate’s interview history into onboarding context for their manager, including working preferences, goals, concerns, and commitments made during hiring.
  </Card>

  <Card title="Interviewer Coaching" icon="graduation-cap" href="/ai-skills/interviewer-coaching">
    Review interview technique against a defined framework and surface specific examples that can support private coaching or team-level training.
  </Card>
</Columns>

### Requires Reports

These skills work across larger sets of conversations and may require access to [Reports](/recruiting-platform/reports), AI fields, or relevant workspace data.

<Columns cols={2}>
  <Card title="Interview Intelligence" icon="chart-line" href="/ai-skills/interview-intelligence">
    Extract and aggregate information explicitly discussed across interviews, calibrate the analysis on real examples, and present the results with appropriate context and limitations.
  </Card>

  <Card title="Role Launch" icon="rocket" href="/ai-skills/role-launch">
    Turn an intake conversation into a draft role brief, interview plan, templates, search criteria, and tracking setup for your team to review and approve.
  </Card>
</Columns>

### Requires Sourcing

These skills use [Sourcing](/recruiting-platform/ai-sourcing), [Application Review](/recruiting-platform/application-review), or talent-market research capabilities.

<Columns cols={2}>
  <Card title="Sourcing Copilot" icon="magnifying-glass" href="/ai-skills/sourcing-copilot">
    Start and refine searches, review candidates against approved role criteria, and prepare provisional shortlists for human review before outreach or ATS actions.
  </Card>

  <Card title="Talent Research" icon="telescope" href="/ai-skills/talent-research">
    Research talent pools, feeder companies, locations, compensation expectations, and market conditions, then combine external findings with relevant themes from your interviews.
  </Card>
</Columns>

## Choose the right skill

| You want to…                                         | Use                                                         |
| ---------------------------------------------------- | ----------------------------------------------------------- |
| Prepare for an upcoming interview                    | [Interview Prep](/ai-skills/interview-prep)                 |
| Organize evidence for a candidate debrief            | [Interview Debrief](/ai-skills/interview-debrief)           |
| Analyze themes across many interviews                | [Interview Intelligence](/ai-skills/interview-intelligence) |
| Prepare onboarding context for a new hire            | [Day-One Brief](/ai-skills/day-one-brief)                   |
| Help an interviewer improve their technique          | [Interviewer Coaching](/ai-skills/interviewer-coaching)     |
| Turn a role intake into a reviewed launch plan       | [Role Launch](/ai-skills/role-launch)                       |
| Find and refine candidates against approved criteria | [Sourcing Copilot](/ai-skills/sourcing-copilot)             |
| Understand a talent market                           | [Talent Research](/ai-skills/talent-research)               |

## Use skills responsibly

<Check>
  **Review before acting.** Skill outputs are drafts. Check important details against the source records before sharing them or using them in a hiring, outreach, or people decision.
</Check>

<Tip>
  **Focus on evidence.** Ask skills to retrieve what was said, organize examples, and identify unanswered questions. Apply your own judgment to what that evidence means.
</Tip>

<Warning>
  **Treat missing information as unknown.** Interview records and public professional profiles may be incomplete. Do not treat an absent detail as evidence that a candidate lacks a qualification.
</Warning>

### Keep people in control

Skills can help prepare information and suggest next steps, but consequential actions remain human-led. In particular:

* Do not use a skill to make a hiring decision or automatically reject a candidate.
* Review provisional candidate notes before applying a verdict, sending outreach, updating an ATS, or adding someone to a sequence.
* Confirm the role criteria and evaluation framework before comparing evidence.
* Keep subjective ratings, overall recommendations, and final decisions with the responsible hiring team.

These principles mirror the guidance in [Best practices for Notes](/account-management/privacy-and-security/ai-best-practices) and [Best practices for Sourcing](/account-management/privacy-and-security/ai-sourcing-best-practices).

### Use factual, job-relevant criteria

Prefer prompts and output fields that focus on relevant experience, examples, requirements, and statements made during the process.

| Prefer                                                                             | Avoid                                        |
| ---------------------------------------------------------------------------------- | -------------------------------------------- |
| “Summarize examples the candidate shared about leading a team.”                    | “Is this person a strong leader?”            |
| “Which approved role criteria are supported by the available profile information?” | “Is this candidate a good fit?”              |
| “Compare the evidence discussed for each competency.”                              | “Rank the candidates from best to worst.”    |
| “Identify unanswered questions for the next interview.”                            | “Infer what the candidate is probably like.” |
| “Refine the search criteria based on this feedback.”                               | “Learn which people we like.”                |

### Handle sensitive information carefully

Only include personal information where it is relevant, appropriate, and available through permitted sources.

* Do not infer protected characteristics, personality traits, emotions, health information, family circumstances, or other sensitive attributes.
* Refer to work authorization only where relevant to legitimate hiring logistics and based on information the candidate or authorized system provided.
* Refer to **compensation expectations** or **voluntarily disclosed compensation information**, rather than implying that Metaview independently knows what a person is paid.
* Exclude compensation and interviewer evaluations from onboarding briefs unless your organization has a separate, appropriate process for sharing that information.
* Redact unrelated sensitive disclosures before sharing outputs more broadly.

### Calibrate aggregate analysis

When using Interview Intelligence or Talent Research:

* Test extraction questions against a representative set of conversations before relying on aggregated results.
* Distinguish information that was not discussed from a negative response.
* Include sample size, time period, scope, and important data limitations alongside charts or conclusions.
* Avoid presenting correlations or patterns as proof of causation.
* Use aggregate or anonymized reporting where named individual data is not necessary.

## Example requests

<AccordionGroup>
  <Accordion title="Prepare for an interview">
    “Prepare me for my interview with Priya. Summarize what has already been discussed, identify relevant unanswered questions, and cite the source conversations.”
  </Accordion>

  <Accordion title="Prepare a candidate debrief">
    “Organize the interview evidence for Amara by our approved competencies. Show supporting examples, contradictions, and unknowns without making the hiring decision.”
  </Accordion>

  <Accordion title="Analyze candidate feedback">
    “Across interviews from the last quarter, summarize the reasons candidates explicitly gave for declining. Calibrate the extraction first and include the sample size and limitations.”
  </Accordion>

  <Accordion title="Refine a sourcing search">
    “Update the search criteria based on this feedback. Treat missing profile information as unknown, and show me the revised criteria before running the next search.”
  </Accordion>
</AccordionGroup>

## Frequently asked questions

<AccordionGroup>
  <Accordion title="Do skills replace prompts?">
    No. A skill provides the reusable workflow, while your prompt tells the AI what you want it to do in the current conversation.
  </Accordion>

  <Accordion title="Do skills make hiring decisions?">
    No. Skills can retrieve and organize evidence, compare it against criteria you provide, and prepare provisional outputs for review. Hiring decisions and recommendations remain with the responsible people in your organization.
  </Accordion>

  <Accordion title="Can I customize a skill?">
    Yes. Skills can be adapted to your organization’s processes, terminology, output formats, evaluation framework, and safeguards. Keep custom instructions factual, job-relevant, and human-reviewed.
  </Accordion>

  <Accordion title="What Metaview data can a skill access?">
    A skill can only use information exposed through the Metaview MCP and available under your existing permissions. It cannot retrieve information that is not present or accessible to you.
  </Accordion>

  <Accordion title="Can a skill change information in Metaview or another system?">
    Some workflows may support actions such as creating a search, updating a candidate record, or preparing an outreach handoff. Consequential actions should require your confirmation and remain subject to the permissions and controls of the connected system.
  </Accordion>

  <Accordion title="Can skills run on a schedule?">
    Scheduling and recurring delivery depend on the capabilities of your AI client and the connections available in your workspace. Installing a skill alone does not create a recurring automation.
  </Accordion>

  <Accordion title="How should I verify an output?">
    Review the cited conversations, transcripts, recordings, scorecards, profiles, or reports used to generate it. Correct speaker labels and incomplete source data before relying on the output.
  </Accordion>
</AccordionGroup>

## Building your own workflows

Skills cover common workflows. For anything they don't — or if you want to build a custom workflow on your own data — see [MCP Use Cases & Workflows](/mcp-guide) for prompting patterns and the Metaview-specific gotchas worth knowing before you start.

## You remain in control

Metaview Skills help you spend less time gathering and organizing information while keeping evaluation and action with your team. Review the evidence, correct errors or missing context, and decide what happens next.