Windsurf Integration
Windsurf can connect directly to Forge Pool through the native MCP server.
Once connected, Windsurf agents can:
- discover Forge capabilities
- inspect execution contracts
- execute deterministic workloads
- retrieve compact results
- analyze uncertainty surfaces
- preserve replay metadata
- assist with Studio-oriented workflows
Forge MCP allows Windsurf to operate as an execution client over uncertainty-aware workloads rather than a traditional code assistant.
Requirements
Before connecting Windsurf, ensure you have:
- a Forge Pool account
- a project
- a project-scoped token
- a Windsurf installation with MCP support enabled
Recommended token type:
fpak_...Project-scoped tokens provide:
- isolated execution context
- deterministic routing
- cleaner governance boundaries
- project-level attribution
- safer billing management
MCP Endpoint
Forge Pool MCP endpoint:
https://api.forgepool.io/mcpAuthentication:
Authorization: Bearer fpak_...Configure Windsurf
Add Forge MCP to your Windsurf MCP configuration.
Example:
{
"mcpServers": {
"forge": {
"url": "https://api.forgepool.io/mcp",
"headers": {
"Authorization": "Bearer fpak_..."
}
}
}
}Replace the example token with your project-scoped token.
Restart Windsurf
After configuration:
- Save the configuration.
- Fully restart Windsurf.
- Reopen the workspace.
Most MCP clients load MCP server configuration during startup.
Verify Connection
Ask Windsurf:
Use Forge MCP.
List available Forge tools.Expected Forge tools:
forge_capabilities_list
forge_capabilities_search
forge_capability_describe
forge_execute
forge_run_status
forge_run_resultIf these tools appear, Forge MCP is connected successfully.
Discovery Workflow
Recommended first prompt:
Use Forge MCP.
Search for insurance loss Monte Carlo capabilities.
Do not execute anything.Expected tool:
forge_capabilities_searchContract Inspection Workflow
After selecting a capability:
Use Forge MCP.
Describe the capability.
Provide required arguments and a minimum valid payload.
Do not execute.Expected tool:
forge_capability_describeAgents should inspect execution contracts before building payloads.
Safe Execution Workflow
Once the contract has been inspected:
Use Forge MCP.
Execute in test mode.
Retrieve compact results.Expected tools:
forge_execute
forge_run_resultRecommended billing mode:
{
"billing": {
"mode": "test"
}
}Example End-to-End Prompt
Use Forge MCP.
1. Search for insurance catastrophe loss capabilities.
2. Select the most relevant capability.
3. Describe the capability.
4. Build a minimum valid payload.
5. Execute in test mode.
6. Retrieve compact results.
7. Summarize expected value, uncertainty, tail risk, and replay metadata.Understanding Results
Forge returns distributions rather than single outputs.
Windsurf should focus on:
- expected value
- median
- percentiles
- uncertainty spread
- tail behavior
- replay metadata
The average alone rarely tells the complete story.
Replay Metadata
Agents should preserve:
- job_id
- trace_id
- request_id
- replay tokens
- execution identifiers
Replay metadata enables:
- deterministic verification
- governance review
- auditability
- scenario comparison
Studio-Oriented Workflows
Forge MCP may expose Studio-oriented tools.
These allow Windsurf to:
- discover templates
- inspect orchestration graphs
- assist workboard design
- prepare template imports
- review execution structure
Agents assist orchestration.
The Forge runtime remains authoritative.
Security Recommendations
Use project-scoped tokens.
Do not:
- commit tokens
- paste tokens into prompts
- reuse production credentials across environments
Recommended:
- separate development tokens
- separate production tokens
- periodic token rotation
Troubleshooting
Forge Tools Do Not Appear
Verify:
https://api.forgepool.io/mcpand:
Authorization: Bearer fpak_...Restart Windsurf after configuration changes.
Authentication Errors
Check:
- token validity
- token scope
- authorization header formatting
- token expiration
Invalid Payload Construction
Ask Windsurf:
Use forge_capability_describe before constructing payloads.
Do not guess profile-specific arguments.The capability contract is authoritative.
Best Practices
Windsurf should:
- search before execution
- inspect contracts before building payloads
- execute in test mode first
- preserve replay metadata
- analyze distributions explicitly
Windsurf should not:
- invent capability identifiers
- guess arguments
- skip contract inspection
- suppress replay metadata
- execute production workloads without approval
Next
Learn the MCP architecture:
Continue with:
→ VS Code
Or return to:
