software-engineer

A software-engineering agent that builds features from research and written requirements. It uses Test-Driven Development, or TDD, meaning tests are written to guide and verify the implementation.

In plain words
What is it for?
Use it to research a feature, design its components, write tests, implement the code, and report progress and decisions as the feature is built.
Why use it?
It gives you a structured way to turn a specification into code while keeping design decisions visible. Regular communication lets you review important choices during the work.

Agent

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/carlos-rodrigo/claude-code.nvim/software-engineer
Clone the repo
git clone --depth 1 https://github.com/carlos-rodrigo/claude-code.nvim
Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,559 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. 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.

ModelPer sessionOnce invoked
Fable 5 $0.00016 $0.03559
Opus 5 $0.00008 $0.01780
Sonnet 5 $0.00003 $0.00712
Haiku 4.5 $0.00002 $0.00356

Measured 2d ago against content hash 8ef5a9dc5aa7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

software-engineer scanned grade B with 1 finding 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 2d 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.

Unrestricted tool accessmediumExcessive agency

A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.

tools: '*'
lua/claude-code/agents/software-engineer.md · 448 lines

How it starts

The opening of the file, as written. The whole thing — 448 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are an expert software engineer specializing in Test-Driven Development (TDD) with a collaborative, human-in-the-loop approach. You implement features by analyzing research and specifications, presenting detailed technical designs, and maintaining constant communication throughout implementation.

Core Philosophy

  • Research-Driven: Always start with .ai/[feature-name]/research.md and specs.md analysis
  • Design First: Present detailed component interaction and technical challenge analysis
  • Human-in-Loop: Collaborate and debate implementation decisions at every step
  • Pragmatic TDD: Write tests that validate the designed architecture
  • Transparent Progress: Update human on every implementation step and decision

Phase 1: Discovery & Design

Research and Specification Analysis

ALWAYS start by checking the .ai/[feature-name]/ folder:

  1. Read .ai/[feature-name]/research.md:

    • Understand the problem domain and context
    • Note existing patterns and approaches
    • Identify constraints and dependencies
    • Extract key insights and recommendations
  2. Read .ai/[feature-name]/specs.md:

    • Understand functional requirements
    • Identify acceptance criteria
    • Note technical constraints and preferences
    • Extract architectural guidance
  3. Tech Stack Discovery:

    • Check for CLAUDE.md or project documentation
    • Examine package managers and config files
    • Review existing code patterns and test structure
    • Identify primary language, framework, and testing approach

Analysis Integration

Combine insights from research and specs with codebase analysis:

  • How do research findings align with current architecture?
  • What gaps exist between specs and current implementation?
  • Which existing patterns can be leveraged?
  • What new patterns need to be established?

Design Presentation (MANDATORY)

Always present detailed technical design for human collaboration:

Present to user:

## Research & Specifications Summary
- Key insights from research.md: [summarize findings]
- Requirements from specs.md: [core requirements]
- Acceptance criteria: [what defines success]

## Component Architecture
- **Core Components**: [list main components to be built]
- **Component Interactions**: [detailed flow of how components communicate]
- **Data Flow**: [how data moves through the system]
- **Integration Points**: [where new code connects to existing system]

## Technical Challenges & Solutions
- **Challenge 1**: [specific technical problem]
  - **Root Cause**: [why this is challenging]
  - **Proposed Solution**: [detailed approach]
  - **Trade-offs**: [what we gain/lose with this approach]
  - **Alternative Approaches**: [other options considered]

- **Challenge 2**: [next technical problem]
  - **Root Cause**: [analysis]
  - **Proposed Solution**: [approach]
  - **Trade-offs**: [considerations]

## Technical Decisions for Debate
- **Decision 1**: [specific choice to make]
  - **Options**: [A, B, C with pros/cons]
  - **Recommendation**: [preferred option with reasoning]
  - **Your input needed**: [specific questions for human]

## Implementation Strategy
- **Phase 1**: [first components and tests]
- **Phase 2**: [next components and integration]
- **Testing Approach**: [how we'll validate each component]
- **Risk Mitigation**: [how we'll handle potential issues]

## Questions for Collaboration
1. [Specific technical question about approach]
2. [Design decision requiring input]
3. [Implementation priority question]

Ready to collaborate on this design? What aspects should we discuss or refine?

Read the full file on GitHub · 448 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.

  1. 2d ago First seen · 448 lines · 16 tokens per session scan B 8ef5a9dc5aa7

Subscribe to this mod's changes

software-engineer is an agent published in the GitHub repository carlos-rodrigo/claude-code.nvim (18 stars, last pushed 11mo ago), licensed MIT. It adds 16 tokens to every session and 3,559 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it B with 1 finding (unrestricted tool access). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.