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

# Vise

> Use the local Vise CLI to guide and validate AI-assisted social.plus SDK integrations.

Vise is a local command-line tool for teams using AI coding agents to integrate social.plus SDKs. It gives the agent a repeatable workflow, grounds the plan in the public docs, writes a local compliance contract, and checks the implementation before you call it done.

The agent still writes the code. Vise wraps that work in guardrails: it grounds the agent in real SDK APIs, checks correctness and completeness, helps align generated UI with your design system, and asks you the calls only a human should make — then keeps checking until the result is green, attested with evidence, intentionally scoped out, or explicitly blocked on your decision.

<Warning>
  **Preview release.** Vise is usable for guided social.plus SDK integrations, but its CLI surface, evidence format, and advisory intelligence may still change as it hardens through more real projects. Treat it as governed AI assistance with strong validation gates — not a substitute for code review, product QA, or runtime verification.
</Warning>

<Info>
  Vise runs on your machine. It fetches public social.plus documentation and SDK version information, but it does not upload your source code, file contents, or search queries.
</Info>

## Vise vs. the docs MCP server

Both tools connect AI coding tools to social.plus, and they work well together.

| Tool                                | Best for                                                                       | Runs where                                |
| ----------------------------------- | ------------------------------------------------------------------------------ | ----------------------------------------- |
| [Docs MCP server](/ai-mcp/overview) | Searching and reading social.plus documentation inside an AI chat              | Hosted at `https://learn.social.plus/mcp` |
| Vise                                | Planning, validating, and sensor-running an integration in your app repository | Local CLI in your project                 |

Use the hosted MCP server when you want an assistant to **find and read documentation**. Use Vise when the assistant is **editing your app** and you want local validation, sensors, and a clear stop condition. Vise can also run as a [local MCP server](/ai-mcp/vise/mcp-server) so MCP hosts call its tools directly.

## Who should use Vise

* **Developer using an AI coding agent** — Install the skill, ask for the SDK feature, answer project-specific questions, and let Vise drive the loop through plan, check, and sensors.
* **Technical lead or reviewer** — Review the `sp-vise/` contract, deterministic findings, attestations, and the `vise check --ci` result before accepting the integration.
* **Partner or solutions engineer** — Use Vise as a repeatable delivery workflow across customer apps, so each integration ends with the same local evidence rather than a prompt transcript.
* **Product or design owner** — Use the advisory design and experience outputs to review whether the generated surface fits the product intent. Correctness and explicit completeness checks remain the gates.

## Install

Install Vise globally:

```sh theme={null}
npm install -g @amityco/social-plus-vise
vise doctor
```

Or run it without a global install:

```sh theme={null}
npx -y -p @amityco/social-plus-vise vise doctor
```

`vise doctor` prints the version, install path, and docs source so you can confirm the setup.

## Install the AI skill

Vise includes a bundled skill that teaches your AI coding tool when to inspect, plan, fetch docs, edit code, validate, and ask you for missing decisions. Install it for your tool:

```sh theme={null}
vise install-skill --target claude     # Claude Code
vise install-skill --target cursor .   # Cursor
vise install-skill --target vscode .   # VS Code
vise install-skill --target copilot .  # GitHub Copilot
vise install-skill --target codex      # OpenAI Codex
```

You can also print the skill and paste it into another coding agent:

```sh theme={null}
vise print-skill
```

## Quickstart

<Steps>
  <Step title="Open your project in your AI coding tool">
    Work inside the app repository where you want to add social.plus SDK features.
  </Step>

  <Step title="Ask for the integration">
    For example: `Add a social feed to this app using the social.plus SDK.`
  </Step>

  <Step title="Let the agent run Vise">
    The skill guides the agent through `vise inspect`, `vise plan`, `vise init`, docs lookup, implementation, `vise check`, and `vise run-sensors`.
  </Step>

  <Step title="Answer project-specific questions">
    Vise asks for values the agent should not invent, such as region, feed target, community ID source, platform scope, or design-system source.
  </Step>

  <Step title="Review the final evidence">
    Done means the local contract is green, attested with evidence, or explicitly blocked on a decision only your team can make.
  </Step>
</Steps>

## Next steps

<CardGroup cols={2}>
  <Card title="How Vise works" icon="gears" href="/ai-mcp/vise/how-it-works">
    The governed loop, what Vise checks, and how to read a check result.
  </Card>

  <Card title="Run Vise as an MCP server" icon="plug" href="/ai-mcp/vise/mcp-server">
    Expose Vise's tools to Claude Code, Cursor, Codex, and VS Code.
  </Card>

  <Card title="CLI reference" icon="terminal" href="/ai-mcp/vise/cli-reference">
    The commands you'll run most, grouped by task.
  </Card>

  <Card title="SDK Quick Start" icon="rocket" href="/social-plus-sdk/getting-started/overview">
    Install and initialize social.plus SDKs for your platform.
  </Card>
</CardGroup>
