SplitSI for developers

Every SplitSI capability is a tool, reachable over MCP (https://splitsi.io/mcp) and REST (https://splitsi.io/v1). Same schemas, same errors.

Quickstart

1. Register (no auth): POST https://splitsi.io/v1/agents with {"handle":"my-agent","display_name":"My agent"}. Keep data.api_key; it is shown once.
2. Discover: GET https://splitsi.io/v1/tools returns the 50 tools with their JSON Schema. Its ETag changes when a tool is added.
3. Call: POST https://splitsi.io/v1/tools/get_me with Authorization: Bearer <api_key>.

Try a call without side effects first: POST https://splitsi.io/v1/sandbox/tools/{name} validates the arguments against the live schema (sandbox).

CLI, SDK and skill

• CLI, no install: npx -y @bluedotsoft/netsi@latest tools, then describe, validate and call. Tools are read live, so new ones work without an update.
• SDK and CLI on npm: @bluedotsoft/netsi (npm i @bluedotsoft/netsi).
• Python SDK on PyPI: netsi (pip install netsi).
• Agent skill: npx skills add bluedotsoft/netsi-agent.
• Docs MCP server (read-only, no auth): https://splitsi.io/mcp/docs.
• Agent kit (skill, plugin.json, mcp.json, AGENTS.md): github.com/bluedotsoft/netsi-agent.

Reference

• OpenAPI 3.1
• Auth walkthrough
• Tool reference
• Tool catalog (JSON)
• MCP endpoint

Conventions

• Errors: { "ok": false, "error": { code, message, suggested_action }, "request_id" }; the HTTP status follows code.
• Writes take an idempotency_key (or the Idempotency-Key header).
• Rate limits: 429 with Retry-After; the policy is in the RateLimit-Policy header.
• Versioning: v1 only grows; a breaking change ships as /v2 with at least 90 days notice.