Aionis Docs

Appearance

Integrations ​

This page maps Aionis into common runtime and framework environments.

How to use this page ​

  1. Pick the runtime or toolchain you already use.
  2. Keep the same Aionis execution pattern across every integration.
  3. Validate the integration with one write, one recall, one policy check, and one replayable ID chain.

Integration pattern ​

  1. Retrieve context before planning.
  2. Apply policy before action.
  3. Persist outcomes and feedback after action.
  4. Keep replay IDs across all steps.

Choose the right path ​

  1. Need a local control plane for tools and experimentation: use MCP and Playground
  2. Need a UI for operators: use Ops Console
  3. Need agent framework orchestration: use LangGraph and OpenWork
  4. Need CLI-driven memory workflows: use OpenClaw

Minimum environment variables ​

bash
export AIONIS_BASE_URL="http://localhost:3001"
export AIONIS_SCOPE="default"
export AIONIS_API_KEY="your_api_key"
# or
export AIONIS_AUTH_BEARER="Bearer <token>"

MCP ​

bash
npm run build
AIONIS_BASE_URL=http://localhost:${PORT:-3001} AIONIS_SCOPE=default node dist/mcp/aionis-mcp.js
npm run -s mcp:smoke

Playground ​

bash
npm --prefix apps/playground install
npm run -s playground:dev

Ops Console ​

bash
npm --prefix apps/ops install
npm run -s ops:dev

LangGraph and OpenWork ​

Recommended order:

  1. recall/context
  2. rules/select
  3. action execution
  4. write/feedback

LangGraph smoke:

bash
npm run -s langgraph:smoke

OpenClaw ​

bash
openclaw plugins install @aionis/openclaw
openclaw aionis-memory bootstrap

Integration checklist ​

  1. Every request carries tenant_id and scope.
  2. Every meaningful write persists at least one recallable node.
  3. Policy-loop calls share the same run_id.
  4. The runtime logs request_id, decision_id, and commit_uri.
  5. At least one recent workflow can be replayed from stored identifiers.