The B2C MCP panel is the workspace control center for connecting an AI app, choosing tool access, generating SFCC guidance, and inspecting requests. Open it with MCP Workspace in the B2C MCP view of the Intellij SFCC sidebar. It opens as the SFCC MCP Workspace editor tab
The sidebar view
The B2C MCP view in the Intellij SFCC sidebar is a compact status card for the workspace server. Its title bar has Pick Agent / IDE Config and Open SFCC MCP Workspace
The status row shows the server state and the action that fits it:
| State | Meaning | Action |
|---|---|---|
| Running | The server answers its health check. The row also shows the number of enabled tools | Stop |
| Stopped | The local server is not running | Start |
| Preparing…, Starting…, Stopping…, Restarting…, Updating context… | An operation is in progress | None until it finishes |
| Not responding | The server did not answer the latest health check | Retry |
| No workspace open | MCP needs an open workspace folder | Open folder |
Running gets an extra line when something needs attention, for example when no SFCC cartridges were detected or the server still runs although intellij-sfcc.mcp.enabled is off. Restart server sits next to the main action while the server runs or does not respond
Below the status are the server URL with Copy URL and the tool access profile, with Customize tool access… to open the tool policy in Settings. The context rows show the Instance of the active connection, its Site and code version, the Workspace, and Activity: requests in flight, the time of the last call, or No agent calls yet
Connect AI… and Pick Agent / IDE Config open the same picker. Choose an AI app, then Copy hosted HTTP config, Copy stdio bridge config, or Open or create project config file. MCP Workspace opens the full panel described below
Without active Intellij SFCC access the view shows SFCC MCP requires an active Intellij SFCC access with the button for your next step (Sign In, Choose a Plan, Renew Subscription or View Assigned Seat, and Open Dashboard when access could not be checked), Sync Access once you are signed in, and Stop local MCP. The Command Palette offers Intellij SFCC: Start SFCC MCP Server, Stop SFCC MCP Server, Restart SFCC MCP Server, Connect AI to SFCC MCP, Open SFCC MCP Workspace, and Open SFCC MCP Activity
Start
Start summarizes the included workspace folders and linked cartridges, the local server, and tool access. Its main action adapts to the server state, such as starting the server or opening Connect AI
The four workflow shortcuts lead to filtered tool lists:
| Shortcut | Tool workflow |
|---|---|
| Read code | Routes, overrides, templates, and cartridge priority |
| Plan & validate | Edit targets and source, XML, or SCAPI validation |
| Inspect an instance | Sites, code versions, remote files, and deployment state |
| Debug | Logs, source mappings, breakpoints, stack frames, and variables |
These shortcuts help you find tools. Review a tool's availability and prerequisites before expecting the agent to use it
Connect AI
Choose an AI app to see its generated configuration, available transports, supported destinations, and verification steps
Choose the destination
Use Save configuration in to select a project or user configuration target. In a multi-root workspace, Choose project file destination selects where project files are written
Choose an available transport
Hosted HTTP connects the AI app directly to the local workspace server. stdio bridge uses a local command that discovers and forwards to that server. Follow the panel's authentication compatibility guidance; a client may require the bridge for the active authentication mode
Save the generated setup
Create config writes a missing, supported target. Open config opens an existing file without modifying it. Use Copy config to review and merge the generated entry yourself when needed
Verify from the AI app
Follow Verify in…, reload the client's MCP configuration, and run Copy safe verification prompt in that client. Use View activity to check the new request and Refresh evidence to update the setup display
The setup summary keeps separate facts visible: whether a file was found, whether this panel copied or created the current setup, transport compatibility, protocol observations, and recent activity. A Stale status means connection or authentication details changed after the recorded setup, so generate the configuration again
For app-specific destinations and skills, see Codex, Claude Code, Cursor, or GitHub Copilot
Tools & Access
Search by task, tool name, or category. Use Understand code, Plan and change, Operate an instance, or Diagnose and debug to narrow the catalog, then select a tool to inspect its details
The table shows Available, Tool, Used for, Action type, and Needs. Those distinctions help explain why a tool is listed but cannot yet run: it may be disabled, require an active connection, or depend on a running debugger
- Change access level opens the profile settings
- Enable safe shown tools applies to the currently displayed safe tools
- Enable all shown… is a separate bulk action
- Reset overrides to profile restores the selected profile's tool choices
- Clear filters restores the full catalog view
Next to the search, Category and four menus narrow the list further:
| Menu | Options |
|---|---|
| Status | Any status, Enabled, Disabled |
| Profile | Any profile, Router Core, Readonly, Operational, Plan, Full |
| Risk | Any risk, Local read-only, Local state change, Remote / open world, Mutation plan, Guarded execute |
| Sort | Catalog order, Name (A→Z), Risk (safe → destructive), Profile (router → full) |
Select a row, or use View selected tool, to open its Tool details, Behavior, and Requirements & activation. When Enable all shown… includes plan or guarded execute tools, a confirmation counts them before Enable Tools applies the change
After changing access, follow the selected client's refresh instructions. A client can keep an older tool list until its MCP connection refreshes
Skills & Guidance
Skill format selects the agent's native output format. Save project files in selects the destination workspace folder. Search by topic, task, trigger, MCP tool, or plugin feature, then inspect a skill and use Add to project. Existing files are preserved; installed skills offer Open installed
Add selected topic generates the chosen topic's skills. Add all skills… generates the full catalog in the selected format. Both show the workspace and destination before writing
The separate Project instruction file control adds optional base guidance for using the workspace MCP. It has its own Instruction format selector and Copy / Add to project actions
See Skills & Guidance for concrete package layouts and an SFRA controller example
Activity
Activity lists recent requests and failures. Start here when an agent cannot discover tools, receives an error, or seems to be using the wrong project
- Run the safe verification prompt from Connect AI
- Find the new request by its timestamp and inspect its result
- Use Show errors only to narrow failures, or clear filters if an expected request is hidden
- Use Copy redacted diagnostics when you need a compact diagnostic report
Request history demonstrates past use. An idle server or an earlier success does not establish that a client is currently connected
The tab is made of these cards:
| Card | What it holds |
|---|---|
| Recent requests | A searchable list filtered by Any activity, Successful, Errors, Guarded / denied, or Slow (≥ 1s), and by tool and transport. Operation timeline lists the plan, validate, execute, and status calls of the same operation under the selected request |
| Selected activity | Status, Method, Tool, Duration, Started At, Transport, Request Id, Agent, and Provider / Model of the request you pick. With payload capture on it also shows Request and Response |
| Diagnostic capture and retention | The Complete payload capture switch, the log folder, file count and size, retention, and Locate files and Clear now… |
| Protocol diagnostics | Copy Inspector UI command and Copy Inspector CLI command to check the server with MCP Inspector |
| Remote SFCC logs | Whether the active connection's logs can be read, and Copy remote-log prompt for your agent |
The Activity tab shows a red badge with the number of recent errors
Open detailed viewer opens the SFCC MCP Activity tab. It charts requests by source over time, lists every logged request page by page, and shows the metadata of the one you select. Its toolbar has Refresh, Open Latest Log, Locate Logs, Export, and Clear Logs. Export saves a redacted sfcc-mcp-activity-redacted.json that leaves out captured payloads and session identifiers
Settings
Settings contains the server controls, authentication configuration, tool policy, client refresh guidance, and activity capture settings. Use the server controls to start, stop, or restart the workspace server when needed
| Section | What you set |
|---|---|
| Server & privacy | Read-only facts: MCP version, transport, server version, connection scope, browser origin checks, and the local diagnostic folder. The server listens only on this computer |
| What the AI is allowed to do | The tool access profile: Connection only, Read this project, Read project and instance, Plan changes, or Full access, with the enabled tool count, and Client refresh |
| Response format | AI provider (Generic, OpenAI / GPT, Claude / Anthropic, Gemini / Google, Kimi / Moonshot), an optional Model name, and Copy provider instructions |
| Protect the local MCP connection | Token authentication with Enable or Disable, Rotate Token, and Copy Env Export |
| Diagnostic logging | Payload capture, captured events, log files and size, with Open Analytics View, Open Latest Log, Locate Logs, and Clear Logs |
Request metadata is always logged on this computer. Complete payload capture under Diagnostic logging also stores full requests and responses, unredacted, for 15 minutes, Until MCP restart, or Indefinite (advanced). Leave it Off unless you are troubleshooting, because captured payloads can contain source code, tokens, and passwords
Keep the workspace open while an agent uses MCP. If you change connection or authentication details, regenerate the client setup and make a new verification request
Troubleshooting
| Symptom | Next check |
|---|---|
| Config exists, but no requests appear | Verify that the AI app loaded this workspace's entry, reload the client, and run the safe verification prompt |
| The agent reports another project | Compare the returned workspace folders with Start and use the configuration generated for this workspace |
| A tool is missing | Clear catalog filters, review Tools & Access, then refresh the client's MCP tool list |
| A remote or debugger tool fails | Inspect its Needs details and confirm the active connection or debug session |
| A generated skill did not change | Existing files are not overwritten. Open the installed file and compare it with the preview before updating it |