Nwflower/dsh-chat-import
Import full-fidelity chat histories from 13 coding agents (Claude Code, Codex, ChatGPT, Cursor, Gemini, opencode, and more) as resumable DeepSeek Harness sessions, with reverse export back to Claude Code.
Install
via npm
Resolves a published tarball rather than cloning the repository, and installs without any extra setup. Swap `web` for your profile name if you run another one.
via GitHub · npm package
Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.
What happened when we ran it
Installed cleanly when we ran it
Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run 14d ago.
Show it in your README
For maintainers: the badge serves this listing's latest sandbox verdict, so a re-run updates it on its own — and it links readers to the full result here.
Due diligence
What it is — Imports chat histories from 19 coding agents into resumable DeepSeek Harness sessions.
Who it is for — When switching between different AI coding agents, this plugin allows continuing conversations seamlessly. Users who only rely on a single AI agent don't need it.
Watch out — Sandbox test passed in a new profile with successful registration. It can execute shell commands. No obvious issues found.
The verdict — I would install it because it supports history import from 19 agents and export back to Claude Code, but only if you frequently switch between tools.
Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.
dsh-chat-import is a DeepSeek Harness plugin that imports conversation histories from external coding agents as full-fidelity, resumable DeepSeek Harness sessions. It reads source histories without rewriting them, creates a fresh session for each imported conversation, and groups sessions by the source cwd. Workspace resolution uses the authoritative project mapping in ~/.claude.json, with slug decoding for Reasonix and a home-directory sandbox guard.
The plugin provides importers for 15 sources named in the README, including Claude Code, Codex, ChatGPT, Cursor, Gemini, opencode, Kimi CLI / Kimi Code, Qoder CLI and local JSONL. It preserves recorded tool calls and results, thinking blocks, titles, models and timestamps where available. ChatGPT branches can become separate sessions, and opencode tool names are mapped to DeepSeek Harness equivalents. Export and handoff operations include export_claude, sync_to_claude, export_codex, export_kimi, export_bundle, restore_bundle, /resume-claude and /resume-codex.
This is intended for developers moving or backing up agent work while retaining resumable context. It is not the right choice for workflows that do not need cross-agent history, or where source formats are unavailable. Export can be lossy, although degradations reports orphan results, skipped injections and skipped attachments. The terminal surface reaches local conversation files, databases, session directories and configuration files because those are the sources and targets of its import, migration and export operations.
DeepSeek Harness loads imported conversations as new sessions rather than modifying the source history or the DeepSeek Harness engine. Source files are read-only. Sessions are grouped into the workspace associated with their source cwd; the plugin consults ~/.claude.json first, applies greedy slug decoding for Reasonix, and uses a home-directory sandbox guard. If the recorded path is not available locally, the source file's directory is used.
The README names one importer per supported source, including import_chatgpt({ branch: 'all' }) for restoring every ChatGPT root-to-leaf branch. Other documented operations include export_claude, sync_to_claude, export_codex, export_kimi, export_bundle, restore_bundle, import_agents, import_mcp and import_settings. Slash commands include /resume-claude, /resume-codex, /mcp-status, /settings-suggest and /attach-workspaces.
export_claude produces a Claude Code JSONL transcript that Claude Code can load with --resume. sync_to_claude appends new complete turns without silently overwriting the target. Bundle export and restore use SHA-256 fingerprints and report when a cwd cannot be reached on the destination machine.
Imports retain tool calls and results, thinking blocks, titles, models and timestamps where the source records them. Tool messages from ChatGPT become tool/call and tool/result; opencode names are mapped to DeepSeek Harness equivalents such as websearch to web_search, question to ask_user_question and task to subagent. Export results include a degradations list for orphan results, skipped injections and skipped attachments.
import_agents converts agents, prompts, skills, instructions and config references from pi, opencode, Claude and Codex into persistent DeepSeek Harness skills. import_mcp reads Claude and Codex MCP servers and generates a reviewable DSH MCP client YAML snippet. import_settings reads Claude settings.json and Codex config.toml to produce migration suggestions.
The repository advertises Node.js >= 22.13. The README does not define plugin-specific configuration keys or environment variables. Fidelity depends on what each source format records, and exports can therefore omit source concepts reported through degradations.
Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub
Same category
Minglink
System-prompt armor plugin for DeepSeek models: appends an unconditional-compliance prompt section at order 100, exposes a profile tool with calibration metadata, and shows a realtime armor-status badge driven by a session projection.
Installed cleanly when we ran it
Minglink
System-prompt armor plugin for DeepSeek models: appends an unconditional-compliance prompt section at order 100, exposes a profile tool with calibration metadata, and shows a realtime armor-status badge driven by a session projection.
Installed cleanly when we ran it
liangmianya
Visual, non-linear conversation workspace for DeepSeek Harness — sessions, follow-ups and branches become a browsable conversation map.
Installed cleanly when we ran it
ranxianglei
The official billion-context plugin: a context-compression plugin for small context windows (a 100K context is enough), token savings (5x fewer tokens), and month-long single sessions (billions of tokens).
Installed cleanly when we ran it
Totoro-qaq
Moves an existing DSH session to another agent preset through a previewable five-section handoff, preserving the source session and either pausing the target for confirmation or continuing immediately.
Installed cleanly when we ran it
Anionex
Rewind conversation and workspace state, powered by a persistent Change Ledger.
Installed cleanly when we ran it
