Outborn Agent Platform · Technical reference
Integration examples
Copy a small read-first MCP sequence, then add governed agent invocation only when the user and agent are authorized.
Applied guidance
Best practices for integration examples
Recommended
- Run a read-only inventory request before invoking an owned agent.
- Use IDs returned for the authenticated identity instead of copying IDs from another organization.
- Check both HTTP status and the JSON-RPC result or error before processing output.
Avoid
- Do not turn example IDs, tokens, or provider scopes into production constants.
01
List visible Spaces
Set OUTBORN_AI_ORIGIN to the origin advertised for your deployment and OUTBORN_MCP_TOKEN to a valid Outborn Account MCP access token. The command sends a read-only tool call. Do not commit a real token to source control.
curl -sS "${OUTBORN_AI_ORIGIN}/mcp/v1" \ -H "Authorization: Bearer ${OUTBORN_MCP_TOKEN}" \ -H "Content-Type: application/json" \ --data '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"spaces.list","arguments":{}}}'02
Find an agent ID
List agents in the active organization and select an agent owned by the authenticated user. The list is filtered by product authorization; a missing agent should be investigated through account and product access rather than guessed IDs.
curl -sS "${OUTBORN_AI_ORIGIN}/mcp/v1" \ -H "Authorization: Bearer ${OUTBORN_MCP_TOKEN}" \ -H "Content-Type: application/json" \ --data '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"agents.list","arguments":{}}}'03
Ask an owned agent
Insert an agent ID returned to this identity. The invocation is a read-only chat request; do not treat the response as confirmation that an external action ran.
curl -sS "${OUTBORN_AI_ORIGIN}/mcp/v1" \ -H "Authorization: Bearer ${OUTBORN_MCP_TOKEN}" \ -H "Content-Type: application/json" \ --data '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"agents.invoke","arguments":{"id":"<agent-id>","message":"What changed in the approved weekly review?"}}}'04
Read the outcome
Check HTTP status and the JSON-RPC error field first. On success, tools/call returns structuredContent.item for a single result or structuredContent.items for a list, plus content text. Keep the request ID with your diagnostic record, while redacting tokens and protected data.
Related documentation