Meshion can now connect directly to OpenAI Codex, Claude, Gemini, Cursor and other MCP-compatible AI clients through the Model Context Protocol.
The new Meshion MCP Server brings Opportunity Intelligence, campaign creation, discovery, human execution and business workflows into the AI-assisted environments where teams already work.
Businesses can analyze a website, prepare campaigns, find relevant conversations or business targets, monitor discovery, review and export opportunities, and coordinate execution—all through natural-language instructions.
This release arrives alongside major improvements to Social Engagement, Target Engagement, website-driven campaign creation, Account Credits, discovery controls, operator workflows and execution integrations.
Together, these capabilities expand Meshion from opportunity discovery into a broader Human + AI operating system for finding and acting on business opportunities.
What Is MCP?
The Model Context Protocol, or MCP, is an open protocol that allows AI clients to connect securely to approved external tools and services.
Without MCP, an AI assistant can explain what to do but may not be able to interact with the systems where the work happens.
With Meshion MCP connected, an authorized AI client can use approved Meshion tools to retrieve campaign information, prepare campaign configurations, start Opportunity Intelligence discovery, monitor runs, review opportunities and coordinate subsequent workflows.
Meshion remains responsible for enforcing access, funding, permissions and workflow rules. Connecting through MCP does not bypass campaign budgets, Account Credits, workspace boundaries, discovery limits, operator controls or submission-review policies.
What Can the Meshion MCP Server Do?
The initial release provides 34 tools across the Meshion workflow, including:
- Website analysis and AI-generated campaign context
- Plain-language campaign Quick Launch
- Products, Services and reusable Targets
- Social Engagement and Target Engagement campaigns
- Opportunity discovery, monitoring and cancellation
- Opportunity review, scoring and filtering
- Complete CSV opportunity exports
- Converting approved opportunities into operator tasks
- Task generation and assignment
- Submission review and revision workflows
- Manual, human-network, webhook and API execution
- Execution integration configuration
- Billing, campaign funding and Account Credit visibility
This makes it possible to move from an initial business objective to a structured campaign and actionable opportunities without manually navigating every part of the Meshion dashboard.
Two Models for Opportunity Intelligence
Meshion Opportunity Intelligence now supports two primary engagement models: Social Engagement and Target Engagement.
Social Engagement
Social Engagement discovers relevant posts, discussions, comments and online conversations where a business may be able to contribute meaningfully.
Campaigns can search supported sources such as LinkedIn, X, Reddit, Facebook, YouTube, Quora, Threads, Medium, Substack, GitHub, blogs and forums.
Discovered opportunities can include:
- The target URL
- Relevance score
- Engagement signals
- Suggested engagement angle
- AI-generated suggested content
This model is useful for finding active conversations around customer problems, industry developments, products, services and topics relevant to a business.
Target Engagement
Target Engagement expands Meshion beyond URL-based social conversations. It is designed to identify people, organizations and other structured business targets, including:
- Individual sales prospects
- Companies and business accounts
- Potential partners
- Investors
- Media contacts
- Recruiting candidates
- Agencies
- Research targets
- Custom opportunity types
Target Engagement opportunities can be enriched, reviewed, filtered and exported. They can also be handled manually, routed into external systems or converted into structured tasks for the Meshion human operator network.
Create Campaigns from a Website or Plain-Language Goal
Businesses can provide a website and let Meshion analyze the product, service, audience, positioning and potential campaign objectives.
Meshion can use this analysis to propose:
- Product, service or target context
- Campaign and context names
- Campaign descriptions
- Discovery instructions
- Required and excluded keywords
- Engagement and opportunity types
- Suggested discovery settings
- Initial opportunity discovery
Basic Quick Launch offers a streamlined path for creating a campaign by answering only a few questions.
Advanced Mode preserves detailed controls for teams that want to configure sources, instructions, scoring, limits and other discovery settings.
When using an AI client, users can ask Meshion to prepare the proposed configuration for review before anything is created or funded. This provides the speed of natural-language campaign setup while preserving business oversight.
From Discovery to Human, Manual, AI or API Execution
Finding an opportunity is only the beginning. Meshion provides several ways to act on approved results:
- Assign work through the Meshion human operator network
- Handle the opportunity manually within the business
- Connect it to an AI-assisted workflow
- Send it to a webhook, CRM or generic API integration
- Export it for external processing
Human-network workflows can include operator matching, task instructions, payouts, suggested content, proof collection, submission review, revision requests, published-link scanning and performance tracking.
This allows businesses to combine AI-assisted research and coordination with human judgment and execution where it matters.
Account Credits and Pay-As-You-Go Usage
Meshion remains a pay-as-you-go platform with no required monthly plan.
Account Credits can support eligible campaign funding, Opportunity Intelligence and API operations. Businesses retain control over discovery limits, candidate-inspection limits, campaign budgets, task payouts and execution spending.
MCP does not introduce a separate billing system. The same Account Credits, campaign balances and API usage controls apply whether an action begins in the Meshion dashboard, through the Business API or from an MCP-compatible AI client.
Compatible AI Clients
Meshion can work with MCP-compatible clients and agent environments, including:
- OpenAI Codex
- Claude Desktop
- Claude Code
- Gemini CLI and Gemini environments with MCP support
- Cursor
- Visual Studio Code environments with MCP support
- Other clients that support remote Streamable HTTP or local stdio MCP servers
Configuration differs by client, but Meshion provides both hosted and local connection options.
Option 1: Connect to Hosted Meshion MCP with OAuth
The simplest connection method is Meshion’s hosted MCP endpoint:
https://mcp.meshion.ai/mcp
To add the hosted server to Codex:
codex mcp add meshion --url https://mcp.meshion.ai/mcp
Then authorize the connection:
codex mcp login meshion
Codex will open Meshion in a browser. Sign in with an approved Business account, select the business organization you want to connect and approve the requested permissions.
The hosted connection uses granular MCP permissions and does not require copying an API key into the AI client.
Hosted MCP is currently intended for approved Meshion Business accounts. Site Admin and Operator accounts cannot authorize the Business MCP toolset.
Other compatible clients can use the same hosted URL when they support remote Streamable HTTP MCP servers and OAuth.
Option 2: Run Meshion MCP Locally Through npm
Businesses that prefer to run the MCP process locally can use the public npm package:
npx -y @meshion/mcp-server
The current release is:
@meshion/mcp-server@0.2.0
First, create a Business API key under Business Settings → API Keys. Choose the MCP granular-access preset or select narrower permissions for the intended workflow.
To add the local npm package to Codex:
codex mcp add meshion-local --env MESHION_API_KEY=msh_live_your_business_api_key --env MESHION_API_BASE_URL=https://api.meshion.ai/api/v1 -- npx -y @meshion/mcp-server
The API key remains on the local machine and is used when the local MCP process calls the Meshion Business API.
A generic MCP client configuration looks like this:
{
"mcpServers": {
"meshion": {
"command": "npx",
"args": ["-y", "@meshion/mcp-server"],
"env": {
"MESHION_API_BASE_URL": "https://api.meshion.ai/api/v1",
"MESHION_API_KEY": "msh_live_your_business_api_key"
}
}
}
}
Example AI-Assisted Workflows
Once Meshion is connected, a user can give instructions such as:
Analyze our website and prepare a Target Engagement campaign for finding companies that may need our services. Show me the proposed configuration before creating it.
Find 25 LinkedIn sales prospects for this campaign, monitor the discovery run and summarize the strongest opportunities.
Find companies associated with crowdfunding and fundraising rather than individual job seekers.
Create a Social Engagement campaign for relevant LinkedIn and X conversations, but do not convert opportunities into tasks until I approve them.
Export every opportunity from this campaign to CSV, not only the opportunities on the currently displayed page.
Show me pending operator submissions, their review signals and whether their published links were recently confirmed live.
Create a webhook integration for sending approved opportunities into our external workflow.
Review our available Account Credits and campaign funding before starting discovery.
Users can also request analysis without authorizing a change:
Analyze the campaign and recommend improved discovery instructions, keywords and minimum score. Do not update the campaign until I approve the recommendations.
This review-first approach provides additional control over actions involving discovery spending, task creation, execution or submission review.
Security and Business Controls
Hosted Meshion MCP uses the OAuth authorization-code flow with PKCE.
Access is:
- Limited to approved Business accounts
- Bound to a selected business organization
- Protected by granular permission scopes
- Revocable
- Automatically expiring
- Isolated from other business organizations
- Exchanged for short-lived internal credentials
OAuth access tokens are stored as hashes by Meshion and are not forwarded directly to the internal Business API.
Local npm connections use the permissions assigned to the Business API key. Businesses can revoke these keys at any time through Meshion Settings.
More Improvements Across Meshion
The MCP release complements a wider set of recent platform improvements:
- Products and Services and Targets libraries
- Reusable campaign context
- Separate Social Engagement and Target Engagement workflows
- Website-driven campaign creation
- Basic and Advanced Quick Launch modes
- Advanced discovery overrides and source controls
- Opportunity deduplication across discovery runs
- Full opportunity CSV exports
- Campaign funding and remaining-balance visibility
- Account Credits
- Improved campaign, opportunity, task and submission pages
- Structured Target Engagement research submissions
- Business-level and global operator blocking controls
- Submission notifications
- Published-link survivability scanning
- Human, manual, webhook and API execution options
- More consistent bulk actions, filters and pagination
Together, these improvements make Meshion a more complete operating system for Opportunity Intelligence and execution—from defining a business objective to discovering, reviewing and acting on the right opportunities.
Get Started with Meshion MCP
Choose the connection method that best fits your environment:
- Hosted OAuth:
https://mcp.meshion.ai/mcp - Local npm package:
@meshion/mcp-server
Hosted OAuth offers the fastest setup and avoids manually copying an API key. The npm package is available for businesses that prefer local execution or API-key authentication.
Connect Meshion to your preferred MCP-compatible AI client and begin discovering, reviewing and executing opportunities through secure, natural-language workflows.
Leave a Reply