dsh-maf
Manifest validMicrosoft Agent Framework (MAF) plugin for DeepSeek Harness: validate, run, and scaffold declarative agents and workflows (YAML).
dsh-maf
Microsoft Agent Framework (MAF) 的 DeepSeek Harness (dsh) 插件 —— 让 dsh 智能体验证、运行和搭建 MAF 的 declarative agents / workflows(YAML 声明式定义)。
独立社区项目,非官方。基于 microsoft/agent-framework(MIT)。 MAF 是 Semantic Kernel 的企业级继任者(v1.0 已发布)。
为什么做
MAF(13k★,Python/.NET)是微软当前的 agent 编排框架,没有 DSH 适配。它的核心是 YAML 声明式 agent/workflow——正好适合 agent 生成、验证和运行:
| 工具 | 行为 |
|---|---|
maf_status | MAF 包版本 + 环境检查 |
maf_validate | 校验 declarative agent YAML(AgentFactory 解析) |
maf_run_agent | 运行 declarative agent(YAML + prompt) |
maf_run_workflow | 运行 declarative workflow sample(main.py) |
maf_scaffold | 生成 agent/workflow YAML 模板 |
前置要求
pip install agent-framework agent-framework-declarative python-dotenv
Agent YAML 的 model.id 常引用环境变量(=Env.OPENAI_CHAT_MODEL_ID),
运行前配置好 .env / 环境变量(provider 密钥)。
安装
# profile 的 cordis.patch.yml
- insert:
- id: maf
name: './src/index.js'
config:
pythonCmd: python
enginePath: scripts/maf.py
pnpm dsh web --patch ./dsh-maf/cordis.patch.yml
然后问 agent:"maf_status 检查环境,maf_validate 校验 ./agents/Assistant.yaml"。
引擎实现
scripts/maf.py 包装 MAF SDK:
AgentFactory(safe_mode=False).create_agent_from_yaml_path(path)→await agent.run(prompt)- workflow 用
python main.py标准 sample 布局(workflow.yaml + main.py)
配置项
| 键 | 默认 | 说明 |
|---|---|---|
pythonCmd | python | Python 解释器 |
enginePath | scripts/maf.py | 引擎脚本路径 |
timeoutMs | 180000 | 单次调用超时(agent 运行可能较慢) |
已知限制
maf_run_agent/maf_run_workflow消耗配置的 provider API 额度- declarative agents API 官方标注为 experimental(可能变动)
- 输出原样返回(含模型生成内容)
目录结构
dsh-maf/
src/index.js # 5 个工具 + 配置
scripts/maf.py # MAF SDK 引擎(status/validate/run/scaffold)
skills/maf/SKILL.md
docs/
cordis.patch.yml
许可
MIT。
Comments
Loading…
Similar plugins
by limuyang2
Multi-agent team collaboration for DeepSeek Harness, with independent models, skills, MCP tools, contexts, and a shared workspace.
★ 46
↓ 428/wk
MIT
TypeScript
Aug 26, 2026
dsh plugin --profile web add @limuyang2/dsh-agent-teamby jiruidai
A model-native meta-agent plugin for DeepSeek Harness that uses the underlying model’s reasoning and planning capabilities to synthesize task-specific workflows at runtime and coordinate tools and sub
★ 7
↓ 85/wk
MIT
TypeScript
Aug 14, 2026
dsh plugin --profile web add dsh-meta-orchestratorby cyanseek
Agent-first DeepSeek Harness plugin intelligence: verify existing plugins, identify missing capabilities, and generate build-ready briefs.
★ 6
MIT
JavaScript
Oct 27, 2026
dsh plugin --profile web add dsh-landscapeby GooDAnDReaDY
Mixture of Agents (MoA) plugin for DeepSeek Harness with /moa slash command, file workspaces, and Live Canvas integration
★ 10
↓ 352/wk
MIT
JavaScript
Oct 23, 2026
dsh plugin --profile web add @goodandready/dsh-moaby Socialist-Sister
Multi-agent collaboration suite for DeepSeek Harness: specialist roster with on-demand dispatch, roundtable, model comparison and a multimodal vision bridge — models via the official provider flow.
★ 4
MIT
TypeScript
Aug 19, 2026
dsh plugin --profile web add dsh-collaborationby luobosibing2
Native DeepSeek Harness (DSH) plugin integrating TypeSafe Jev as a System One decision layer for agent selection, supervision, corrections, and approvals.
★ 213
MIT
JavaScript
Sep 28, 2026
dsh plugin --profile web add @dsh-jev/plugin