dsh-codex-import

Many developers have built up Codex configurations, hooks, Skills, and AGENTS.md instruction files over time and may want to move to DeepSeek Harness (DSH). The hard part is figuring out which pieces are directly compatible, which need rework, and which simply will not work—all without risking credential leaks, accidental hook execution, or breaking the permission model. dsh-codex-import exists to answer that "what actually fits?" question before you commit to a migration.

It is a strictly read-only compatibility scanner that walks your user-level and project-level config.toml files, hooks.json, standalone Skill directories, and standard AGENTS.md / AGENTS.override.md files, then classifies every item as READY, DISCOVERED, NEEDS_REVIEW, SKIPPED_UNSUPPORTED, or UNSUPPORTED with a brief reason. It also identifies which of the five synchronous command-hook types (PreToolUse, PostToolUse, SessionStart, UserPromptSubmit, Stop) the current DSH hooks-codex bridge can actually run. Throughout the scan, nothing is written to disk, no profile is modified, no hooks are executed, and no chat history is touched. The output deliberately contains only categories, statuses, counts, and generic reasons—never config values, command bodies, Skill text, or absolute paths.

It is best suited for individual developers or small teams evaluating a Codex-to-DSH transition: run one scan, get a structured compatibility list, and decide your next steps with confidence. If you maintain both toolchains side by side, it also serves as a practical difference memo, flagging which Skills need rewriting and which hooks are temporarily unsupported, so you can plan around security-semantic gaps instead of discovering them mid-migration.