Existing Architecture Integration Assessment
Running Claude Code, Codex, Grok, or Agents built on mainstream SDKs? Evaluate integration pathways into ReadyForAI.
Existing CLI and SDK Agents connect through RuntimeClasses without a rewrite
- Vendor RuntimeClass hosts Claude Code / Codex / Grok CLIs and Claude Agent SDK / OpenAI Agents SDK processes
- Native uses the Go Native SDK; Sidecar is the A2A / stdio / ACP bridge, with Google ADK on Sidecar HTTP/A2A
- Context files are injected without changing Agent source; then phase into the governance loop
- Skill ingest uses LarkScout's three-layer gate; HarnessServer SkillScan can run before Workspace deploy. Both are static checks, not a runtime sandbox by default
Native RuntimeClass
Connect in-house high-performance Agents with the NodalOS Go Native SDK.
Vendor RuntimeClass
Host vendor CLIs (Claude Code / Codex / Grok) and official SDK processes (Claude Agent SDK / OpenAI Agents SDK).
Sidecar RuntimeClass
Bridge existing Agents via A2A / stdio / ACP; Google ADK uses Sidecar HTTP/A2A. Not a sidecar container process.
Basic RuntimeClass
Run new scenarios directly using NodalOS built-in general-purpose Agents.
- 01Select RuntimeClass
- 02Register to NodalOS
- 03Connect Observability
- 04Enable Policy Hooks
- 05Onboard to AULO Surface
Start with a low-risk Agent to validate the complete path (Registration → Observability → Approval) before batch-migrating the rest. Big-bang cutovers are never required.
By role
Technology leaders
Multi-form RuntimeClasses + A2A / ACP / MCP deliver open interoperability without vendor lock-in
Compliance & risk
Fully traceable integration process; HITL approval checkpoints activate during the pilot phase
Business leaders
Protects prior Agent investments by enabling phased, low-risk migration