Agent platform research corpus
This corpus is the frozen research input behind the platform's agent-architecture decisions: an industry study of agent products plus the running record of the design discussion that followed it. Where a conclusion here differs from an accepted record in the ADR index, the ADR is authoritative.
Method
The research prompt is preserved so the shared scope and evidence rules behind each product dossier remain reproducible.
Product dossiers
- ADK + A2A
- AWS AgentCore Harness
- Bedrock AgentCore
- Claude Code Agent SDK
- Claude Managed Agents
- Cloudflare Agents
- CrewAI
- DeepSeek Harness
- Devin
- Google Agent Executor (AX)
- Guild
- Hermes Agent
- IronClaw (NEAR AI)
- Jido
- kagent
- LangChain Deep Agents
- LangGraph Platform
- LangSmith Managed Deep Agents
- LiteLLM
- Netclaw
- OpenAI Agents API
- OpenAI Agents SDK
- OpenClaw
- OpenComputer
- Standard Agents, a later version-pinned study of its portable package and runtime specification, with a primary-source ledger
- Vercel
Contract impact
- Contract impact: OpenAI Agents API, the dossier findings read back against
proto/trogonai/agents/agents/v1alpha1andproto/trogonai/session/sessions/v1alpha1. Analysis only; every proposed change is gated on an ADR. - Contract impact: Standard Agents 0.2.0, a later comparison with the inspected platform checkout, separating executable infrastructure, draft contracts, semantic mismatches and adoption evidence.
Synthesis
- Synthesis: what the industry means by "agent", the cross-product convergence and divergence analysis drawn from the dossiers above.
Decision record
- Agent Service: design questions and answers, the running Q&A (Q1 through Q26) that turned the synthesis into directional design decisions, including the mistakes surfaced along the way and where the design is heading next.
Later research on the same providers
The provider agent contracts corpus re-studied three of the products above, plus xAI, against a narrower question: how each provider designs sessions, agents, and vault secrets. It goes deeper on credentials than the dossiers here and does not revise them.
