Blog · Comparison
"It doesn't want to be the next Codex" — DSH vs Codex
August 14, 2026 · dshbase
Calling DeepSeek Harness "a DeepSeek version of Codex" is the first thing people say — and the thing the project most pushes back against.
Agent presets, not scenarios
Codex-class products separate modes by scenario (daily/office, office/code). DSH instead uses Agent presets: the exact set of plugins an agent runs with — tools, prompts, and capabilities. You copy one, tweak it, or build a new one, including non-coding presets like a writing workspace.
Rebuildable sessions
Codex sessions are opaque. DSH sessions are append-only event logs — system prompts, reasoning, tool calls, permission changes all become events. When an agent fails or loops, you trace the exact step. An observability win Codex doesn't expose.
The plugin model
Where Codex lets users customize Skills and MCP but keeps the rest baked in, DSH makes everything a plugin — the UI, tools, workflow. You assemble an agent, not rent one.
Model flexibility
DSH supports directory and custom providers — you could run GLM inside it. Codex is tied to OpenAI models.
Practically
- Codex-class: polished, scenario-driven, minimal configuration.
- DSH: assemble presets, rebuild sessions, audit every step, extend the agent itself.