# plori-ai/plori [Health: Active]

**Category:** 🤖 Coding Agents  
**Repository:** https://github.com/plori-ai/plori  
**GitHub Stars:** 0  
**Views:** 6  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/plori-ai-plori

## Description
Give your AI agent its own cloud computer. Create and drive hosted plori agents (persistent disk, real tools, memory that survives between sessions) over a remote MCP server: invoke an agent and read its reply, manage the human-in-the-loop queue, and schedule deferred runs. OAuth 2.1 sign-in or API key.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "plori": {
    "command": "npx",
    "args": ["-y","plori-mcp"]
  }
}
```

## Documentation

## What plori-ai/plori MCP server does

The plori-ai/plori MCP server gives an MCP client access to hosted agents running in plori.ai. Each agent has its own persistent machine, disk, tools, and memory, so state can remain available between sessions. The server is intended for driving those agents from an AI client or another MCP-compatible application rather than operating files on the developer's computer.

Available operations span the agent lifecycle and its work. You can list, inspect, create, and delete agents; start runs and retrieve their replies; inspect or cancel active work; and handle questions that require human input. The service also supports deferred runs and multi-step workflows. Account tools expose credit balances, usage, disk information, and the configured third-party connections without returning tokens or client secrets.

## How it works

The service is hosted and exposes a streamable HTTP MCP endpoint at `https://api.plori.ai/mcp`. There is no local MCP process to build or launch for clients that support this transport. Interactive clients authenticate through a browser using OAuth 2.1 on first connection. Headless environments can send a plori API key as a bearer token instead.

Runs can be blocking, allowing the client to wait for a result, or fire-and-forget. Workflow operations include reading definitions, editing drafts with compare-and-swap behavior, starting executions, and checking execution status and per-step data. Creating or running agents consumes credits, while list operations, result retrieval, and balance reads are free according to the supplied documentation.

## Setup and configuration

For the plori-ai/plori MCP server, configure a remote HTTP MCP connection with the endpoint `https://api.plori.ai/mcp`. Cursor provides an add-server flow and Cline has a dedicated installation guide. Claude Code and VS Code examples are also provided in the project documentation, as is a Codex CLI setup that performs OAuth login.

A client that only supports stdio can use the separately published `plori-mcp` npm wrapper, which forwards requests to the same remote endpoint. That wrapper is not the MCP service itself. For CI or other noninteractive environments, pass an API key in the `Authorization: Bearer ...` header. Keys are created from Dashboard → Settings on a registered plori account.

## Tools and capabilities

The documented tool groups include:

- **Agents:** list, inspect, create, and delete agents.
- **Runs:** invoke agents, read results, list runs, retrieve past results, and cancel in-flight work.
- **Human-in-the-loop:** list pending agent questions and submit answers.
- **Scheduling:** arrange a run for a later time.
- **Workflows:** list, inspect, edit, create, execute, and monitor workflows, including version and step data.
- **Account reads:** check credits, usage, disk details, and connected OAuth providers.

The agent platform's router selects a model for each task. Workflow execution is a real, billed operation, while account and historical read operations do not consume credits as described in the README.

## Limitations and notes

This connection does not provide the cloud agent with access to the user's local files. Shell commands, files, and disk data belong to the agent's hosted environment. The service requires a registered account for API-key creation, and agent creation or execution can incur credit usage. OAuth access can be revoked in the MCP client's settings, while API keys can be revoked from the plori dashboard.

The repository is an integration front door for the hosted product; the remote MCP implementation is served by plori.ai rather than run from this repository. Clients must support streamable HTTP directly or use an appropriate bridge for stdio-only environments.

_Full upstream README: https://allmcps.com/mcp/plori-ai-plori/readme_