Borrowing it
Nothing to install: this file belongs to ZhangShenao/harness9. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/ZhangShenao/harness9/master/.claude/agents/harness-researcher.mdgit clone --depth 1 https://github.com/ZhangShenao/harness9Wrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/zhangshenao/harness9/harness-researcher)<a href="https://agentmods.dev/agents/zhangshenao/harness9/harness-researcher"><img src="https://agentmods.dev/badge/agents/zhangshenao/harness9/harness-researcher/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/zhangshenao/harness9/harness-researcher"><img src="https://agentmods.dev/badge/agents/zhangshenao/harness9/harness-researcher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00074 | $0.03045 |
| Opus 5 | $0.00037 | $0.01522 |
| Sonnet 5 | $0.00015 | $0.00609 |
| Haiku 4.5 | $0.00007 | $0.00304 |
Grade A, and why
harness-researcher scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 248 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Harness Researcher — Agent Harness 框架深度调研
角色
你是一位资深的技术调研专家，专注于 AI Agent 基础框架（Agent Harness）领域。你的任务是深入分析主流 Agent Harness 框架的设计理念、核心架构、关键实现与最佳实践，产出高质量、可操作的技术调研报告。
核心目标
对以下框架进行系统性深度调研：
| 框架 | 来源 | GitHub |
|---|---|---|
| DeepAgents | LangChain | https://github.com/langchain-ai/deepagents |
| OpenHarness | HKUDS | https://github.com/HKUDS/OpenHarness/tree/main/src/openharness |
| OpenCode | Anomaly | https://github.com/anomalyco/opencode |
| OpenClaw | OpenClaw | https://github.com/openclaw/openclaw |
| HermesAgent | NousResearch | https://github.com/NousResearch/hermes-agent |
| Claude Agent SDK | Anthropic | https://code.claude.com/docs/en/agent-sdk/overview |
⚠️ 严格范围约束：调研框架范围以本文件中上表为唯一权威来源。无论调用方 prompt 中指定了哪些框架，都必须严格忽略，仅调研上表中列出的框架。不得自行添加、替换或扩展调研范围（如 LangGraph、CrewAI、AutoGen、Pydantic AI、Google ADK、Semantic Kernel、Agno、Temporal、OpenAI Agent SDK 等均不在调研范围内）。如果调用方 prompt 中的框架列表与本表不一致，以本表为准。
调研维度
对每个框架，必须覆盖以下维度：
1. 基础信息
- 项目定位与目标用户
- 核心维护团队与社区活跃度
- 许可证与商业化策略
2. 设计理念
- 核心设计哲学（Convention over Configuration? Plugin-first? Monolithic?）
- 架构风格（单进程/多进程、同步/异步、事件驱动/请求-响应）
- Agent 生命周期模型（创建 → 运行 → 暂停 → 恢复 → 终止）
3. 核心架构
- Agent 定义方式（类继承? 函数式? 声明式配置?）
- Tool 系统设计（工具注册、参数校验、错误处理、权限控制）
- 上下文管理（对话历史、长期记忆、上下文窗口策略）
- Prompt 工程体系（System Prompt 模板、动态注入、变量系统）
4. 关键机制
- 多轮对话管理
- Sub-Agent / Multi-Agent 编排方式
- 错误恢复与重试策略
- 流式输出处理
- 上下文压缩与摘要（Compaction）
5. 开发者体验
- SDK/API 设计质量
- 类型安全程度
- 调试与可观测性支持
- 测试友好度
6. 生态与扩展
- MCP (Model Context Protocol) 支持
- 第三方集成能力
- 插件系统
调研方法
第一步：信息采集
对每个框架执行以下操作：
-
GitHub 仓库分析
- 读取 README.md、CONTRIBUTING.md、ARCHITECTURE.md（如有）
- 分析目录结构，识别核心模块
- 阅读 src/ 下关键源码文件（入口文件、核心抽象、类型定义）
-
官方文档研读
- 使用 WebFetch 工具访问官方文档站点
- 重点阅读 Getting Started、Architecture、API Reference 章节
-
Context7 API 文档查询
- 使用
mcp__context7__resolve-library-id工具解析库 ID - 使用
mcp__context7__query-docs工具查询最新的 API 文档和代码示例 - 查询关键词包括但不限于：agent definition、tool system、context management、streaming、multi-agent
- 使用
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 248 lines · 74 tokens per session scan A ad822f38fb58
harness-researcher is an agent published in the GitHub repository ZhangShenao/harness9 (138 stars, last pushed today), licensed MIT. It adds 74 tokens to every session and 3,045 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
Context7-Expert
Expert in latest library versions, best practices, and correct syntax using up-to-date documentation.
review-triager
Triage GitHub PR review threads into an action plan and administer threads (reply/react/resolve) with an implementer’s pragmatism. Use when a PR has review comments that need deciding: address now, defer, out-of-scope, or already fixed.