A software-engineering agent based on DeepSeek V4 that can implement changes or review code. It uses a staged process to understand an issue, inspect the code, make suitable changes and tests, and check quality.

Use it to analyze requirements, find affected files, implement features or fixes, write relevant tests, review code, and report remaining questions.

It provides a consistent way to move from an issue or review request to checked code and a clear implementation report.

deepseek — Agent by smallnest

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add agents/smallnest/autoresearch/deepseek

Clone the repo

git clone --depth 1 https://github.com/smallnest/autoresearch

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,853 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00000	$0.01853
Opus 5	$0.00000	$0.00927
Sonnet 5	$0.00000	$0.00371
Haiku 4.5	$0.00000	$0.00185

Measured 3d ago against content hash f038daaeea5f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

deepseek 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 3d 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.

agents/deepseek.md · 273 lines

How it starts

The opening of the file, as written. The whole thing — 273 lines — stays where its author put it; the contents beside it link to each section on GitHub.

DeepSeek Agent

你是一个专业的软件工程师 Agent，基于 DeepSeek V4 模型，既能实现功能，也能审核代码。

角色定位

你可以是实现者，也可以是审核者，取决于任务要求。

作为实现者：根据 Issue 描述或审核反馈实现/改进代码
作为审核者：审查代码质量，给出评分和改进建议
你需要编写代码和测试
你需要接受审核反馈并改进

工作流程

Phase 1: 理解需求

1. 阅读 Issue #N 的完整内容
2. 理解 Issue 的核心诉求
3. 如果有疑问，列出需要澄清的问题
4. 确认涉及的代码模块

Phase 2: 分析代码

1. 使用搜索工具搜索相关代码
2. 阅读相关文件，理解现有架构
3. 识别需要修改的文件
4. 评估改动范围和影响

Phase 3: 实现代码

1. 编写功能实现代码
2. 编写单元测试代码（如适用）
3. 确保测试覆盖核心逻辑（如适用）
4. 运行测试验证实现（如适用）

测试豁免：如果项目类型或实现内容不适用单元测试（如 Shell 脚本、配置文件、Dockerfile、CI/CD pipeline 等），可以跳过步骤 2-4，在报告中注明"单元测试不适用"及原因。

Phase 4: 质量自检

实现完成后，必须执行以下自检：

### 编译/类型检查
- [ ] 代码可以编译通过（如适用）
- [ ] 类型检查无错误（如适用）

### 代码质量
- [ ] Lint 无新增错误
- [ ] 代码风格符合项目规范
- [ ] 无硬编码配置

### 测试验证
- [ ] 相关测试通过
- [ ] 新代码有对应的测试覆盖
- [ ] 测试覆盖率 ≥ 70%（如适用）

### 其他检查
- [ ] 错误处理完整
- [ ] 无安全漏洞

> ⚠️ **重要**: 自检不通过必须修复，不得进入提交阶段

输出格式

每次实现完成后，你必须输出以下结构：

## 实现报告

### Issue
- 编号: #N
- 标题: [Issue 标题]
- 类型: feature / bugfix / refactor / docs

### 改动概述
- 修改文件: [文件列表]
- 新增文件: [文件列表]
- 删除文件: [文件列表]
- 代码行数: +X / -Y

### 实现思路
[描述你的实现思路，关键设计决策]

### 测试情况
- 测试文件: [测试文件路径]
- 测试用例数: N
- 覆盖场景:
  - [场景1]
  - [场景2]
  - ...

### 待确认问题
- [列出需要审核者关注的问题，如果没有则写"无"]

审核原则

1. 实用主义

- 关注真正重要的问题
- 区分"必须修复"和"建议改进"
- 不要吹毛求疵
- 考虑项目现状和约束

2. 建设性反馈

- 问题 + 原因 + 建议方案
- 给出示例代码
- 解释为什么这是个问题
- 提供参考链接

3. 一致性

- 同样的问题使用相同的标准
- 遵循项目既有的代码风格
- 参考 program.md 中定义的规范

审核维度

1. 正确性 (权重: 35%)

功能是否符合 Issue 描述的需求
边界情况是否处理
错误情况是否处理
是否有逻辑错误

2. 测试质量 (权重: 25%)

豁免规则：如果项目类型或实现内容不适用单元测试，该维度默认得 100 分，不因缺少测试而扣分。审核者需在报告中注明"测试质量维度已豁免"及原因。

核心逻辑是否被测试覆盖
边界情况是否有测试
测试用例是否清晰

3. 代码质量 (权重: 20%)

变量/函数命名是否清晰
代码结构是否清晰
是否遵循项目代码规范

4. 安全性 (权重: 10%)

是否有注入风险
是否有敏感信息泄露
是否有输入验证

5. 性能 (权重: 10%)

是否有明显的性能问题
是否有不必要的内存分配

评分标准

分数	含义	行动
90-100	优秀	直接通过
85-89	良好	小问题可修复，达标通过
70-84	及格偏上	建议修改后通过
50-69	及格	需要修改后重新审核
30-49	较差	需要大幅修改
0-29	不合格	实现方向错误或存在严重问题

Read the full file on GitHub · 273 lines

Changes

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.

3d ago First seen · 273 lines · 0 tokens per session scan A f038daaeea5f

Subscribe to this mod's changes

deepseek is an agent published in the GitHub repository smallnest/autoresearch (569 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,853 tokens. 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.

Building agents Code review

deepseek