Mission Squad Docs

Appearance

Mission Squad Platform — Getting Started ​

Audience: Builders who prefer UI over raw API calls. This page gives you the fastest path to a working setup and maps the UI to the corresponding API endpoints.

Deployment Modes ​

  • Cloud‑hosted
    • Managed by Mission Squad; minimal setup.
  • Customer‑hosted

Concepts at a glance ​

  • Providers: secure, per‑account API credentials for model vendors (OpenAI, Google, Anthropic, Groq, OpenAI‑compatible APIs, or your own externally accessible API).
  • Models: named configurations you create per provider model (temperature, max tokens, etc.).
  • Agents: task‑specific wrappers around a model + system prompt, optional Tools (MCP), optional RAG collections.
  • MCP: Model Context Protocol servers that expose tools (functions) agents can call.
  • Prompt Studio: scratchpad to run models/agents, observe tool events, and save prompts back to agents.
  • Embeddings: vector stores (knowledge bases) created from your files for RAG.
  • Workflows: multi‑agent orchestration (primary + helper agents).
  • Schedules: recurring runs of an agent/workflow with delivery (email; Slack if configured).

First‑run checklist (≈10 minutes) ​

  1. Add a Provider → Providers tab → enter API key for, e.g., openai.
  2. Discover & Save a Model → Models tab → select provider & provider model → give it a friendly name (e.g., gpt-4.1-mini-factual).
  3. Install/Enable Tools (optional) → MCP tab → connect servers (e.g., webtools, github) and confirm they show "connected" with listed tool schemas.
  4. Create an Agent → Agents tab → bind to your named model, paste a system prompt, optionally enable Tools and pick functions; optionally attach Embeddings collections.
  5. Test in Prompt Studio → choose Target Model or Agent → toggle Streaming ON to watch tool events; iterate prompt.
  6. Schedule it (optional) → Schedules tab → pick agent, paste the prompt, set frequency/time, set recipients → enable.

API parity (developer reference):

  • Providers: GET/POST /v1/core/providers, POST /v1/core/add/provider, POST /v1/core/delete/provider — see Providers
  • Models: POST /v1/core/models, POST /v1/core/add/model, POST /v1/core/delete/model, GET /v1/models — see Models
  • Agents: GET /v1/core/agents, POST /v1/core/add/agent, POST /v1/core/delete/agent — see Agents
  • Tools/Servers inventory: GET /v1/core/tools, GET /v1/core/servers — see Core Utilities
  • Chat runs: POST /v1/chat/completions (OpenAI‑compatible; model can be a model name, one of your published agent names, or shared/<ownerUsername>/<slug>) — see Chat Completions
  • Embeddings & files: /v1/files, /v1/vector_stores — see Files and Vector Stores
  • Workflows: POST /v1/core/agent-workflow — see Core Utilities

Top navigation — Header with tabs for Providers, Models, Agents, Workflows, Schedules, Embeddings, Prompt Studio, MCP. The active tab is highlighted.

  • Providers — add credentials and view each provider’s Supported Models snapshot.
  • Models — create named configurations per provider model; adjust generation params.
  • Agents — define task prompts, bind to a model, and enable Tools (MCP) and Embeddings (RAG).
  • Workflows — compose primary + helper agents.
  • Schedules — configure recurring runs and delivery.
  • Embeddings — create/search vector stores; upload files.
  • Prompt Studio — run quick tests, stream events, and save prompts back to an agent.
  • MCP (Tools) — manage servers; inspect available tools and their input schemas.

Next steps ​