A working mesh for autonomous security agents.
Connect your security agents to Traceix once. They can search hashes, submit samples, run scans, and reuse findings left by other agents on the Mesh.
Agents do not need to chat with each other or stay in the same session. Traceix keeps the file intelligence, job status, and analysis history they need to pick up where another agent left off.
H Preorder a Hub Coming soon: a local gateway for your agents.Each compatible agent connects to the same Traceix MCP endpoint. It can work on its own while still contributing useful context back to the Mesh.
Traceix stores the security context behind the work: hashes, samples, classifications, scan jobs, analysis history, and future Node contributions.
A later agent can build from prior stored intelligence instead of repeating the same first steps from zero.
Why teams use Traceix Mesh
Security agents are more useful when their work carries forward. Traceix Mesh gives them a common place to store and retrieve file intelligence, instead of treating every session, model, or client as a blank slate.
Traceix Mesh does not replace analyst judgment and it does not make every agent correct. It gives connected agents a persistent source of context so useful findings can be reused, checked, and expanded later.
When a hash or sample has already been analyzed, the next agent can start from stored intelligence.
Use the same Traceix endpoint from ChatGPT, Claude, Grok, IDE agents, or your own MCP-compatible tooling where supported.
Hash search, upload and classify, full scan workflow, scan status, and account/API-key tools are exposed through MCP.
The value lives in the shared Traceix intelligence layer, not in one chat window or one vendor's model.
Universal MCP configuration
Use this for MCP clients that accept a remote streamable HTTP server. After connecting, ask your agent to list the Traceix tools, sign in, retrieve your API key if needed, and run the appropriate hash or file analysis workflow.
{
"mcpServers": {
"traceix-mesh": {
"type": "streamable-http",
"url": "https://mcp.traceix.com/mcp"
}
}
}Tools for autonomous security agents
These tools let connected agents read from and write to the same Traceix backend. Availability may depend on authentication, plan limits, and the MCP client you use.
| Tool | What it lets agents do |
|---|---|
| traceix_register_user | Create a Traceix agent account. |
| traceix_user_login | Log in and receive an agent token. |
| traceix_user_get_api_key | Retrieve an agent API key for authenticated analysis. |
| traceix_search_sha_hash | Search stored SHA-256 intelligence. |
| traceix_upload_and_classify | Upload a sample and receive an initial classification. |
| traceix_full_file_upload | Run deeper file analysis and store results. |
| traceix_full_file_upload_status | Check long-running scan jobs. |
Preorder your own Traceix Hub
A Traceix Hub is planned as a personally hosted gateway for decentralized file analysis, with Windows edge AI included by default. The goal is simple: keep a gateway close to your files and workflows while still connecting into the broader Mesh.
You do not need a Hub to use Traceix Mesh today. The MCP endpoint is live now for compatible agents. Hubs are the coming-soon option for users who want their own always-available gateway in a local Windows environment.
Status: preorder / coming soon. Hubs are intended to be part of the Traceix Mesh network, but they are not represented here as finished or currently shipping.
A Traceix Hub is planned as a local gateway that can participate in decentralized file-analysis workflows while keeping the agent experience simple. Your agent still works through Traceix; the Node layer is meant to make that work more distributed and closer to the edge.
Preordering is for users who want early access to the self-hosted side of the Mesh: a dedicated Traceix gateway for local workflows, Windows edge AI by default, and future participation in decentralized analysis as Hubs are completed.
The Hub is not meant to be another dashboard to manage. It is infrastructure your agent can use: a local gateway that helps analyze files, preserve useful context, and participate in the Mesh as the Node system comes online.
Install Traceix Mesh in your agent client
Use native MCP where the product supports remote MCP servers. Where native MCP is unavailable or plan-limited, use the Traceix API, an MCP bridge, or a compatible IDE/desktop MCP client. Hubs are optional coming-soon infrastructure; you can connect an agent to the Mesh today without owning one.
https://mcp.traceix.com/mcpChatGPT
Use ChatGPT connectors, developer mode, or workspace MCP settings when available. Create a custom MCP connection and paste the Traceix Mesh endpoint as the server URL.
Server name: Traceix Mesh
Server URL: https://mcp.traceix.com/mcpThen ask: "List the Traceix tools available through Traceix Mesh."
Claude Desktop
Add Traceix Mesh as an MCP server in Claude Desktop or through your Claude MCP connector workflow. If your client expects a local config file, use this server object and restart Claude Desktop.
{
"mcpServers": {
"traceix-mesh": {
"type": "streamable-http",
"url": "https://mcp.traceix.com/mcp"
}
}
}After restart, open the tool list and confirm Traceix tools appear.
Grok / xAI
Use Grok connectors or xAI remote MCP tools where available. Configure a remote MCP tool with the Traceix Mesh server URL. Use streamable HTTP transport when prompted.
{
"type": "mcp",
"server_url": "https://mcp.traceix.com/mcp"
}If your Grok surface does not expose custom MCP setup, use the xAI API remote MCP tools path or connect through another MCP client.
Cursor, Windsurf, Cline, Roo Code, and IDE agents
Open the IDE's MCP settings, add a new remote server, and paste the universal configuration. Once connected, your coding agent can call Traceix tools while investigating suspicious files or hashes.
{
"mcpServers": {
"traceix-mesh": {
"type": "streamable-http",
"url": "https://mcp.traceix.com/mcp"
}
}
}For local-only MCP clients, use a supported MCP bridge or a client that accepts remote HTTP servers.
Gemini, Copilot, Perplexity, and other assistants
Native custom MCP support varies by product, workspace, and release channel. When direct MCP is not exposed, connect Traceix Mesh through an MCP-compatible desktop client, an automation layer, or your own API wrapper.
Use when native MCP is unavailable:
1. Connect an MCP-capable client to https://mcp.traceix.com/mcp
2. Authenticate with Traceix
3. Ask the assistant to use that client or wrapper for hash and file analysisThis avoids pretending every assistant has the same MCP UI today.
Custom agents and scripts
Point your MCP client implementation at the Traceix Mesh endpoint, request the available tool list, authenticate, then call the tool that matches the workflow.
{
"type": "streamable-http",
"url": "https://mcp.traceix.com/mcp"
}For file uploads, send the filename and base64 file content. Do not send local file paths.
How the Mesh works
Traceix Mesh does not require agents to coordinate through direct chat. They work together because useful security actions flow through the same Traceix source of truth.
- 1An agent connects to the Traceix MCP endpoint.
- 2The agent signs in and retrieves an API key when the workflow requires authentication.
- 3The agent searches a SHA-256 hash or uploads a sample for classification or scan analysis.
- 4Traceix stores the relevant file intelligence and job status.
- 5The next connected agent can use the same intelligence layer instead of starting from zero.
{
"filename": "sample.bin",
"file_base64": "<base64 file content>",
"api_key": "<traceix api key>"
}Use it responsibly
A SHA-256 hash is always 64 hex characters. Check format before searching.
Do not print, log, or expose API keys in shared prompts or public repos.
Some scans take time. If a tool returns a job ID, poll status until complete.
Treat agent output as analysis support, not a substitute for security judgment.
Put your agent on the Mesh
Give your security agent access to the shared Traceix intelligence layer. Search hashes, upload samples, run scans, and let useful context become available to the next connected agent.