模型与 AI 能力

Provider-neutral LLM service interface for the DeepSeek Harness

Shared anonymous user identity for DeepSeek Harness telemetry and feedback correlation

Shared LLM generation policy for DeepSeek Harness session-title providers

File-backed settings provider (settings.yaml) for the DeepSeek Harness

DeepSeek MCP Server - Model Context Protocol server for DeepSeek API

All-user-messages LLM provider plugin for DeepSeek Harness session titles

Background job registry (ctx.jobs) for the DeepSeek Harness — shared ids, owner isolation,

Models settings and official-DeepSeek first-run routing over one live provider/settings/cr

Process-wide outbound HTTP proxy policy for DeepSeek Harness: resolve it from the launch e

DeepSeek Harness plugin: keeps model requests under a gateway's request-body ceiling by re

A self-improving agent. Tested on DeepSeek, Qwen, Kimi, GLM models. Supports MCP, Skills.

Client app for ChatGPT, Gemini, Claude, Kimi, DeepSeek, Grok, Nani !?, Perplexity, Genspar

Vision tools (analyze_image, locate_element) for DeepSeek Harness: local ollama, DSH subag

Use your DeepSeek, OpenCode Go, or GLM (Z.ai) API subscription with Claude Desktop — local

DSH 插件 th1nk-er/dsh-plugin-http-proxy

基于 DeepSeek Harness 的自进化智能体框架：40 个 Cordis 插件、4 套 preset，含记忆、反思、门禁、度量与可审计的进化闭环。不是一条命令装完。需克隆

让 DSH Agent 通过 Harness² 的 Propose–Probe–Compose 循环，从真实执行证据中持续改进提示词、技能、脚本与领域经验库。

DSH 输入区的按会话执行档位：OVERDRIVE 开关只对当前会话开启 Turbo，极速/极限两档替换 agent 的规划与验证策略、并按档位调整请求的 reasoningEff

DSH drawio ai plugin

每轮回复结束时用独立审查模型审核输出，判定不合格则引导 agent 修复，每轮最多追问 2 次；可注入 SOP 文件夹标准作为额外审核维度。

Model-facing vision tool for DeepSeek Harness: analyze images via any OpenAI-compatible vi

DeepSeek Harness plugin: compresses verbose tool error payloads before they reach model co

Manage the local llama.cpp Qwen3.5-4B from a DS-Harness Settings page: start/stop, launch