Recipes

由 AgentRunDB 真实运行支撑的可复现配置,而非 AI 生成的教程。

Develop a Custom Plugin for DeepSeek Harness (dsh)

Step-by-step guide to writing and publishing a modular Cordis-based plugin for DeepSeek Harness.

0 次 AgentRunDB 真实运行支撑

Attach GitHub MCP Server to Codex CLI

Give OpenAI Codex CLI direct repository inspection, PR creation, and issue tracking capabilities via Model Context Protocol.

2 次 AgentRunDB 真实运行支撑 · 成功率 50%

Connect OpenCode to DeepSeek V3

Complete guide to configuring OpenCode TUI with DeepSeek V3 for fast terminal development.

2 次 AgentRunDB 真实运行支撑 · 成功率 50%

Use DeepSeek R1 Reasoning Model with Claude Code

How to route complex reasoning and architectural planning through DeepSeek R1 while maintaining agent stability.

5 次 AgentRunDB 真实运行支撑 · 成功率 80%

DeepSeek Harness plugins: the dsh-plugin ecosystem

Everything is a Plugin — how dsh plugins are discovered, and what the ecosystem looks like today.

0 次 AgentRunDB 真实运行支撑

Install and run DeepSeek Harness (dsh)

Run the official DeepSeek agent harness in one command — Web UI on localhost:3080. From the official README.

0 次 AgentRunDB 真实运行支撑

Fix DeepSeek tool-calling issues

The three failure modes we actually hit in runs — and the fix for each. Based on real AgentRunDB evidence.

4 次 AgentRunDB 真实运行支撑 · 成功率 75%

DeepSeek + Agent Skills

Which agents honor Agent Skills when running on DeepSeek — current verified state and workarounds.

4 次 AgentRunDB 真实运行支撑 · 成功率 75%

DeepSeek + MCP setup

Attach MCP servers (e.g. GitHub) to an agent that runs on DeepSeek. Configuration verified with Claude Code + GitHub MCP.

4 次 AgentRunDB 真实运行支撑 · 成功率 75%

Use DeepSeek with OpenCode

OpenCode has first-class custom-provider support — set the base URL and model in one config block.

0 次 AgentRunDB 真实运行支撑

Use DeepSeek with Codex

Configure Codex CLI against a DeepSeek OpenAI-compatible endpoint. Source-level recipe — AgentRunDB runs pending.

0 次 AgentRunDB 真实运行支撑

Use DeepSeek with Claude Code

Point Claude Code at the DeepSeek API with two environment variables. Verified by 4 AgentRunDB runs at 75% success.

4 次 AgentRunDB 真实运行支撑 · 成功率 75%