Platform Overview
The Grimoire Platform is a cloud API that builds on the open source engine, adding natural language intents, policy enforcement, session management, and sandboxed execution.
What the Platform Adds
The open source CLI lets you write and run spells locally. The Platform wraps that engine in an API and adds capabilities that matter for teams and integrations:
- Natural language intents — describe what you want in plain English, get a validated action plan back
- Policy enforcement — define rules that automatically approve, flag, or reject actions
- Clarification sessions — when intent is ambiguous, the API asks follow-up questions instead of guessing
- Sandboxed execution — strategy planning runs in isolated environments
- Audit trail — every run, decision, and policy evaluation is persisted
Feature Comparison
| Capability | Open Source CLI | Platform API |
|---|---|---|
| Write and compile spells | Yes | Yes |
| Simulate/preview | Yes | Yes |
| Onchain execution | Yes | Via executor |
| Natural language intents | — | Yes |
| Policy enforcement | — | Yes |
| Clarification sessions | — | Yes |
| Sandboxed execution | — | Yes |
| Multi-user API keys | — | Yes |
| Audit persistence | Local history | Managed database |
How It Works
Intent → Plan → Validate → Preview → (optional) Execute- Intent — you send natural language or a structured action plan
- Plan — the platform translates intent into a Grimoire spell
- Validate — policies are evaluated against the plan
- Preview — the spell runs in simulation to verify outcomes
- Execute — if approved, the plan is committed
If the intent is unclear, the platform enters a clarification session — asking targeted questions until the plan is unambiguous.
API Surface
| Endpoint | Purpose |
|---|---|
POST /v1/validate | Validate an action plan against policies |
POST /v1/run | Create an intent session from natural language |
POST /v1/session/{id} | Submit clarification answers |
GET /v1/session/{id} | Get current session state |
| Policy CRUD endpoints | Manage partner policies |
Next Steps
- Getting Started — set up authentication and make your first API call
- Validate Actions — validate action plans
- Run Intents — use natural language to create strategies