# Avowex — the trust layer for AI agents

> Wrap a risky agent action in a human-in-the-loop check: escalate → resolve →
> resume. Free tier: 500 resolved transactions/month, Mode B single review.

## Integrate
- [AI integration guide](/console/avowex-ai-integration.md): full SDK (Python + TypeScript) and REST reference, a copy-paste recipe, and a ready prompt. Hand this whole file to your AI assistant.
- SDKs: `pip install avowex` · `npm install @avowex/sdk` · MCP server: `npx @avowex/mcp`
- API base: https://api.avowex.com  ·  Site: https://www.avowex.com

## Core
- POST /v1/orgs — create a free workspace, get an avx_ API key
- POST /v1/action-types — register a gated action + policy
- POST /v1/escalations — escalate (pause the action)
- GET  /v1/escalations/{id}/decision — read the verdict
- POST /v1/escalations/{id}/resume — resume the agent
- GET  /v1/plan — plan, usage, remaining