DeepSeek Harness Plugins · Workflow
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A plugin for DeepSeek Harness that provides cross-session verification, claims, and a decision ledger to prevent agent sessions from repeating, contradicting, or deadlocking each other.
A community plugin for DeepSeek Harness that enhances workflow by routing tasks to native capabilities, verifying readiness, and safely activating reviewed paths.
dsh-tool-chaos is a deterministic fault injection tool for testing the resilience of DeepSeek Harness tool pipelines, generating evidence reports through baseline, dry-run, and fault injection.
A community plugin for DeepSeek Harness that exposes longgraph skills for long-horizon agent work, enabling prompt library and workflow orchestration.
A beginner-friendly plugin for DeepSeek Harness that demonstrates tool registration, structured output, and multi-language greeting generation.
dsh-vision-bridge adds image understanding to text-first models in DeepSeek Harness, supporting multiple vision providers with a glasses toggle for enabling image analysis.
A curated bundle of verified MCP servers for DeepSeek Harness (dsh), enabling easy integration of battle-tested servers into agent workflows.
A plugin for managing skills in DeepSeek Harness, supporting listing, searching, deleting, recovering, and version updates.
Turn DeepSeek Harness into a virtual product development team, guiding users from an idea to a shippable product through an automated pipeline without requiring technical expertise.
The dsh-task-control plugin adds task pause, resume, and cancel capabilities to DSH Web, supporting safe and forced pausing.
This plugin integrates open-mcp-apps applications into DeepSeek Harness, providing each app with its own workspace and conversation interface for real-time interaction and inline rendering.
A personal workbench plugin for DeepSeek Harness that integrates calendar, task list, and AI assistant, featuring natural language task intake, AI execution with user acceptance, and local data storage.