Start with the product you are shipping.
Compose and ship the runtime graph: providers, Agent Loop, tools, policy, Sessions, and interfaces.
Give Claude a governed coding workspace and extend it with CLAUDE.md, Skills, MCP, hooks, agents, and plugins.
Use a repository-aware coding agent across local and hosted workflows with AGENTS.md, Skills, MCP, plugins, sandboxing, and approvals.
Use a coding agent to build DSH extensions, or expose shared services through MCP.
Do not translate names literally.
Carry a reviewed summary and artifacts, not a copied transcript.
Rebuild discovery, activation, policy, and removal for the target.
Test filesystem, network, process, escalation, and admin policy separately.
Choose Claude Code or Codex when the coding-agent product is the desired boundary. Evaluate DeepSeek Harness when your deliverable is the inspectable, replaceable control plane itself.
Run one controlled evaluation.
same repository snapshot
same bounded task
explicit model/provider
recorded tools and policy
restart and recovery testCompare completion evidence, effect containment, durable state, extension lifecycle, latency, provider cost, and operator time. Feature-name parity is not an acceptance test.
Primary evidence.
- DeepSeek Harness rc.8 architecture
- Claude Code extension overview
- Claude Code permissions
- Official Codex repository
- Codex configuration schema
Keep the complete decision matrix.
The canonical guide includes seven product dimensions, migration traps, a practical selection test, ten evaluation gates, and official sources for all three products.
Read the complete comparison