multi-agent-systems

Native DSH rooms for connecting independent Sessions and provider-backed AI members—without bundled roles or scenarios.

Durable dependency-graph task factory for DeepSeek Harness: recurring Agent work, safe checkout lanes, first-class queues, Triage, and artifacts.

Execution-time drift firewall for long-running DeepSeek Harness agents. Real-Harness tests: unsafe stale mutations 12/12 native -> 0/12; valid controls 7/7 both; post-SIGKILL unsafe continuation 2/2 -> 0/2.

Multi-agent orchestration and LLM model routing for DeepSeek Harness (DSH): semantic AI agent profiles, exact model routes, declarative teams and strategies, and bounded subagent delegation - a TypeScript plugin that adds no second runtime.

Goal-driven multi-agent orchestration for DeepSeek Harness — bounded DAG scheduling, nested delegation, typed evidence, and durable trajectory UI.

Agent skills for running research projects end-to-end using a plain-file epic→issues→comments tracker.

LLM loop that reasons in a dense technical register — mechanical gates + adversarial judge + reverse-translation for humans. Self-upgrading. Standalone or as a DSH plugin.

DeepSeek-PA（DPA）是基于 DeepSeek Harness（DSH）构建的非官方模块化桌面工作台，提供独立的 Chat 与 Agent 空间、插件与 Skill 扩展、全局主题，以及可监督的多智能体集群项目。Unofficial modular desktop workspace for DeepSeek Harness (DSH), featuring separate Chat and Agent views, plugins, Skills, global themes, and supervised multi-agent projects.