Showing the last successful snapshot, data dated 2026-11-12
The DSH Plugin
Leaderboard
Search a plugin, check install verification, then copy the exact command.
DeepSeek Harness plugin for Reactive Resume: bridges your resumes and job applications into a Harness session over MCP.
iPolloWork Design Studio and its curated design templates as a native DeepSeek Harness conversation view.
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).
AgentRQ task manager for DeepSeek Harness: create, manage, and auto-pull AgentRQ tasks without leaving the harness.
Visual, non-linear conversation workspace for DeepSeek Harness — sessions, follow-ups and branches become a browsable conversation map.
A desktop-native BigFish companion driven by DeepSeek Harness session events.
Adds an archived-sessions page in Settings to search, restore, and delete archived DeepSeek Harness sessions by workspace.
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.
Conversation rewind: one click back to any earlier user message, in the same window — no new branch, no leftover copy; the target message is refilled into the composer, with optional workspace-file restore — the full Claude Code /rewind semantics.
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.
Rewind conversation and workspace state, powered by a persistent Change Ledger.
Inline-edit and recall your own user messages in dsh web — lazily and seamlessly, with a version pager and per-session draft persistence.
Full session management for the DSH web UI: delete (with trash/restore/purge), restore archived sessions, activity stats, continue/pause, and open log folders. No harness changes.
DeepSeek Harness session and workspace manager —— Workspaces: archive and visibility. Sessions: move, migrate presets, delete, favorite, review later, prioritize, tag, and annotate. Batch operations.
Filters tool outputs, prunes stale context and discloses skills on demand for DSH 0.2.1-alpha.1.
Delete one complete closed turn from a DeepSeek Harness session without deleting the session.
Execution maze for DSH sessions: the agent's real path — main advance, failed or dead-end detours, backtracks, request retries and subagent branches — drawn on one wall-clock timeline, with three tracks under each lane sharing that axis (tool-call density, token pulse, context pressure) and a deterministic analysis panel below it (tool result matrix, failure-recovery chains, duration distribution); watch it grow live in a session tab, or upload up to 5 session logs for a same-axis compare.
Cross-session messaging across one or all workspaces: session_list finds top-level sessions by recency or title, session_send_message steers or queues attributed messages (cold-resuming stored sessions when needed), session_create opens titled top-level sessions with an initial task, and a client renderer displays relayed messages as user-side chat bubbles naming the sender session.
DSH web plugin: select conversation text -> ask in a right-side panel -> a dedicated follow-up session (❓<topic>) in the same workspace, without interrupting the main conversation. Runs on DSH's own right sidebar, with no extra dependencies.
One-command install of the curated first-party DSH plugin suite (plugin-manager store.
Primary DSH entry for using ChatGPT with local code, Agent Helm engineering capabilities, and native DSH sessions.
Adds /btw for one-off, context-aware side questions without tool execution, with independent answer bubbles that can be copied, collapsed, or closed without adding answers to the main conversation.
Session-aware sandbox escalation compatibility plugin for DeepSeek Harness.
Recall any user message: rolls the conversation (official session fork) and workspace files (per-message shadow git snapshots) back to before it, with a diff-preview confirmation panel.
Delete DSH sessions from the web UI and desktop client: header danger button and session-row menu item with a risk-consent dialog; host endpoint and agent tool remove the session log, projection cache, and workspace accounting.
Using DSH and plugins
What is DeepSeek Harness?
DeepSeek Harness, or DSH, is DeepSeek AI's open-source agent harness. It uses an everything-is-a-plugin architecture powered by Cordis and is currently a developer preview.
How do I run DSH?
Install Node.js, then run npx @deepseek-ai/dsh web. The Web UI starts at http://127.0.0.1:3080 by default.
What is a DSH plugin bundle?
A distributable plugin bundle declares dsh.bundle.patch in its package.json. The patch contributes one configuration layer to a DSH profile.
How do I install a plugin?
Use dsh plugin --profile <profile> add <package-or-github-spec>, replacing <profile> with the DSH profile you want to use, such as web. Compatible plugin pages show the exact installation command. Official runtime: DeepSeek Harness.
How is DSH compatibility checked?
CI first checks dsh.bundle.patch, then runs the exact dsh plugin add command in an isolated profile. Successful installs rank first; manifest-confirmed bundles remain available with a lower compatibility score while verification is pending.
Are DSH Web's built-in plugins ranked?
No. The Web profile already ships with @deepseek-ai/dsh-base and @deepseek-ai/dsh-web-app. Both are official bundles and are excluded from the third-party leaderboard.
Is it safe to install plugins?
Verification confirms the manifest format, not that third-party code is safe. Review the source, pin a commit, and only allow install-time build scripts for repositories you trust.
How is this leaderboard ranked?
Each GitHub repository occupies one rank, even when it contains multiple bundles. Only repositories that pass an isolated dsh plugin add run receive a rank; GitHub stars decide their order.