DeepSeek Harness Plugins · Workflow
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Integrates the macOS open-record-replay toolchain into DeepSeek Harness, enabling agents to learn desktop workflows through recorded user demonstrations.
Transforms DSH's binary plan approval into a precise, multi-comment review workflow featuring exact text annotation, consolidated feedback, and a built-in read-only AI Q&A assistant.
Exposes DeepSeek Harness's agent runtime capabilities as a standard MCP server, enabling external clients to efficiently delegate and execute structured coding tasks.
A durable control layer for long-running development workflows in DeepSeek Harness, offering persistent DAGs, dependency management, crash recovery, and automated verification gates.
Unifies AI skills from DSH, Codex, Claude, and GitHub in a single panel, featuring one-click installs, hot toggles, and instant activation.
A closed-loop Agent trajectory control plane for DeepSeek Harness that tracks task continuity, information gain, and verification debt to enforce strict benchmark validation and execution governance.
A testing engineering skill framework for AI agents that empowers them to act as senior QA engineers, executing end-to-end testing pipelines from requirement analysis to test case generation and automation.
Turns any content into native promotional materials for 17 platforms, preserving core ideas and original links.
This plugin bridges Claude Code's memory, skills, and configuration into DeepSeek Harness with zero migration and full compatibility.
A DeepSeek Harness plugin that controls Android devices via ADB and drives tasks through an observe-decide-act-verify loop.
A runtime requirement drift guard that maintains a durable baseline and silently monitors long-running agents, only interrupting to confirm material directional changes.
Automatically retries failed model calls with backup models to keep long tasks running, while providing an orchestrator tool to split complex work into observable, parallel sub-agent workflows.