DSH Marketplace

dsh-interconnect

Chinesezjc/dsh-interconnect

Cross-instance message and event handoff between DSH instances via an interconnect server.

373TypeScriptMITSource

Install

Add dsh-interconnect to DeepSeek Harness

Install into

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 todayon DeepSeek Harness 0.2.2-rc.5.

Show it in your README

install verified — dshmarketplace

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

Before you install dsh-interconnect

  • Source of record: Chinesezjc/dsh-interconnect — present in the community registry that DSH's own plugin market installs from.
  • Licensed under MIT.

Security

What installing dsh-interconnect runs

Build scripts

No build script was blocked in our run.

Source: our sandbox install on DeepSeek Harness 0.2.2-rc.5, today.

Install hooks

package.json declares no preinstall, install or postinstall script.

Source: package.json on GitHub, read at our last sync. prepare is not recorded; it runs only on a GitHub install, and shows up under build scripts when pnpm blocks it.

README mentions

shell commands (bash, exec, spawn …)

Source: a keyword match on the README — a hint about what the plugin touches, not an analysis of its code.

Not an audit. These are the record of one install and two text scans; nobody has read this plugin's code on your behalf. Plugins run with your agent's permissions.

The AI take

What it is — DSH instances exchange messages and events via an interconnect server.

Who it is for — When cross-instance message and event pushing is needed. If only a single DSH instance is used, this plugin can be skipped.

Watch out — Sandbox test passes with installation in a new profile and registration into the profile. Static check shows it will execute shell commands.

The verdict — I would install it since it allows bidirectional event pushing between DSH instances, although wakeup is disabled by default.

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.

What dsh-interconnect does

Abstract relay connecting paired chambers with message threads and branching event streams.

dsh-interconnect is a DeepSeek Harness plugin that forwards messages and events between DeepSeek Harness instances through an interconnect server. It adds a host-level interconnect service and model-visible tool-interconnect tools. The service exposes HTTP send, list and ping endpoints under /interconnect/*, plus a /interconnect/link WebSocket for bidirectional event streaming, heartbeats and reconnects. Incoming events are emitted as interconnect/event.

DeepSeek Harness instances authenticate requests with the shared DSH_INTERCONNECT_TOKEN bearer credential; authentication fails closed when the credential is absent. interconnect_send targets a session, while interconnect_list discovers live sessions and interconnect_ping checks a peer's liveness and identity. Messages can use followup, steer or inject delivery. With explicit resume: true, a peer may wake a persisted offline session, which can start a billable agent turn with that session's configured tools.

This is intended for deployments that need cross-machine session addressing, event fan-out or agent-to-agent handoff. It is not a substitute for local session messaging, and it is a poor fit when remote delivery or waking sessions is unnecessary. The resume option is off by default and can be refused by the receiving instance. Because it can wake an agent with its full tool set, including a terminal capability when that session has one, dsh-interconnect reaches the permissions and execution surfaces granted to the receiving agent.

dsh-interconnect documentation

Configuring dsh-interconnect

DeepSeek Harness loads both plugin entries at host composition level. Configure the interconnect entry as follows:

Key Default Description
instanceId dsh Diagnostic identity returned by ping, send and list; it is not used for routing.
requestTimeoutMs 10000 Outbound request timeout, capped at 60000.
peers [] Event fan-out targets configured at startup; subscribe() can add targets at runtime.
delivery followup Default for inbound messages without an explicit delivery mode.
allowResume true Whether remote senders may wake persisted offline sessions.

The shared credential is not part of this config. DeepSeek Harness reads DSH_INTERCONNECT_TOKEN from credentials such as .credentials.yaml. Both instances must use the same token. Missing credentials produce 403 responses.

Commands and endpoints

DeepSeek Harness exposes interconnect_send, interconnect_list and interconnect_ping as model-visible tools. interconnect_send takes a peer baseUrl, a target sessionId and message text, with optional delivery and resume. interconnect_list returns live session IDs and may include title and status; interconnect_ping reports peer liveness and identity.

The host service provides HTTP /interconnect/send, /interconnect/list and /interconnect/ping endpoints, and the /interconnect/link WebSocket endpoint for event delivery.

How dsh-interconnect behaves

followup queues an independent next turn, steer inserts at the nearest step boundary or starts a turn when idle, and inject writes context without waking an idle agent. A send result reports session-not-live, unreachable, resume-refused or resume-failed when delivery fails. Delivered messages are persisted with source.kind set to plugin; the source metadata is not included in model context.

Known limits

Only live sessions are listed and directly reachable. resume needs DeepSeek Harness's configured agent lookup; deployments without that lookup, including headless or profiles without api-proxy, fall back to session-not-live. Sessions using an old disk format cannot be resumed and return resume-failed.

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

Alternatives to dsh-interconnect

billion-context

ranxianglei

7.9k

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

npm packageTypeScript2d ago

AI review

billion-context

What it is — Shrinks session context to fit a ~100K window; the repo claims about 5× fewer tokens and month-long uninterrupted threads.

Who it is for — You are in DeepSeek Harness, keeping one agent session alive for weeks while token use and the context window start colliding. If your environment blocks plugins that run shell commands, skip it; if a prefix-cache drop would never send you to /acp, skip it too.

Watch out — A clean-profile install registered with the harness. The repo puts compression overhead at ≤2% and a healthy prefix-cache hit rate at 96–99%; sustained drops are meant to be inspected with /acp or /acp-cache. The 5× saving and billion-token month were not checked in this test. The plugin does run shell commands.

The verdict — I’d attach it to a disposable long thread and let /acp plus the invoice vote—nothing else.

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.Read the source
1Details
3.6k

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

GitHub sourceJavaScript25d ago

AI review

dsh-infinite-gen-4

What it is — The blurb describes a DeepSeek system-prompt armor plugin; the README is an IG5 reverse-engineering workbench.

Who it is for — If you run DeepSeek models as DSH agents and need a compliance block pinned at order 100 plus a session-projection armor badge, the blurb is the match. If you came for static analysis, native debug, and traffic capture on a sample, that writeup is under dsh-infinite-gen-5 and should not be treated as this plugin.

Watch out — The sandbox only showed a fresh-profile install that the harness registered; prompt mutation and the badge were not exercised. No install-time failure showed up.

The verdict — I will not let a registration-only plugin touch the system prompt of a profile I actually work in.

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.Read the source
Details
3.3k

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

GitHub sourceJavaScript1mo ago

AI review

dsh-infinite-gen-3

What it is — A DeepSeek-oriented system-prompt armor for DSH: it appends an unconditional-compliance block at order 100, exposes a profile tool that carries calibration metadata, and renders a live armor-status badge from a session projection.

Who it is for — You run DeepSeek models inside DSH on long, cooperation-heavy sessions where refusals break the work, and you want the armor state and calibration metadata visible on the current session. If you already own the system prompt through the DSH profile and do not look at badges or calibration data, this layer is surplus.

Watch out — Sandbox install succeeded on a fresh profile and the harness registered the plugin. The short repo blurb describes armor; the README we have describes a different product—IG5 (dsh-infinite-gen-5), a reverse-engineering workbench. Confirm the tree and the actual prompt text before you treat either description as the thing you are installing.

The verdict — Source registration is cheap; putting an unread compliance block into a daily profile is not—read the injected section in a throwaway profile first.

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.Read the source
Details

dscode

qiz029

998

A DeepSeek coding agent harness: persistent shell, Ultra subagents, auto approval, Chrome MCP and session telemetry

Did not install when we ran it

GitHub sourceJavaScript23d ago

AI review

dscode

What it is — A DeepSeek terminal coding agent with a persistent shell, file edits, and sessions on the same machine that can hand off work and review diffs.

Who it is for — You are on macOS or Linux, driving DeepSeek against a repo with builds and tests running, and you juggle more than one change set at a time. Skip it if you only want a small plugin inside a DSH profile you already run, not a replacement terminal agent.

Watch out — After a completed install, DSH did not register it into the profile, so it is not available as a plugin. You must supply an API key or token. It will run shell commands as part of the coding loop.

The verdict — Leave a working DSH install untouched. Come back only if profile registration starts succeeding, not because the feature table got longer.

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.Read the source
2Source

dsh-synapse

liangmianya

728

Visual, non-linear conversation workspace for DeepSeek Harness — sessions, follow-ups and branches become a browsable conversation map.

Installed cleanly when we ran it

npm packageJavaScript16d ago

AI review

dsh-synapse

What it is — dsh-synapse provides a visual non-linear conversation workspace for DeepSeek Harness, turning sessions, follow-ups and branches into a browsable map.

Who it is for — When managing multi-turn conversations with branches and follow-ups in the DeepSeek Harness web profile, this tool helps organize them visually. Users who prefer native DSH interfaces for linear conversations or do not use web profiles may not require the additional mapping tool.

Watch out — Installation requires pnpm and DeepSeek Harness >=0.1.7-rc.2 with Node.js >=22.19.0. Sandbox testing confirmed successful installation and registration in a new profile. No obvious issues found.

The verdict — I would install it because it preserves native session forks for branching while adding a visual map, but only under the web profile requirement.

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.Read the source
Details

Adds an archived-sessions page in Settings to search, restore, and delete archived DeepSeek Harness sessions by workspace.

Installed cleanly when we ran it

npm packageJavaScriptyesterday

AI review

dsh-archive-manager

What it is — Adds an archived-sessions page in DeepSeek Harness Settings for searching, restoring, and deleting archived sessions by workspace.

Who it is for — When handling multiple workspace conversations in DeepSeek Harness and needing to archive inactive sessions to keep the main interface clean, this plugin can be used. It is not suitable for users who only manage a small number of conversations in a single workspace, because it is designed for bulk search and restore operations.

Watch out — After installation, the Archived sessions page appears in Settings. Sandbox testing passed and the plugin registered into the profile. Large conversations may run slowly due to lack of persistent indexing.

The verdict — I would install it because it enables diagnosis and repair of archived sessions, which offers a specific advantage in long-conversation workflows.

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.Read the source
Details