Choose an Integration
Pick the integration based on where your Agent loop already runs.
| If you use | Start here | Why |
|---|---|---|
| Node or TypeScript Agent | TypeScript SDK | Full application integration once you own the host loop. |
| Any service language | Raw HTTP | Call product APIs directly. |
| Agents that read workspace files | AIFS | Expose governed context, receipts, blocked memory, and snapshots as read-only project files. |
| Cursor, Zcode, or another MCP client | MCP | Tool bridge for route-safe execution context when a native adapter is not installed. |
| Claude Code | Claude Code | Native hooks inject Aionis context before prompts and record execution evidence after tool use. |
| Planner/worker/verifier/reviewer | Multi-Agent Hosts | Share governed execution state across roles. |
| Mem0, Zep, Supermemory, vector DB, markdown, logs | External Memory | Use Aionis as a governance layer. |
Product-first Choice
| Goal | Product |
|---|---|
| Continue long-running Agent work | Execution Memory |
| Govern third-party memory candidates | Memory Firewall |
| Replay what happened after an incident | Agent Flight Recorder |
| Make loop-engineered Agents remember safely | Loop Engineering |
When in doubt, start with SDK and compact Agent context.