code-survey

Surveys codebase dimensions relevant to a requirement before design work begins.

2|Updated Jun 8, 2026
One-click install
npx skills add https://github.com/HACK-WU/skills --skill code-survey-hack-wu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-survey
Source: https://github.com/HACK-WU/skills/tree/main/skills/code-survey
Command: npx skills add https://github.com/HACK-WU/skills --skill code-survey-hack-wu

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Before designing a feature, engineers often lack context about the existing codebase—its style, architecture, API conventions, and data storage patterns. This Skill performs an on-demand, requirement-scoped survey of the codebase so design decisions stay consistent with existing code. ## Core Features & Use Cases - On-demand dimension selection: Chooses only relevant dimensions from 13 options (code style, architecture, API conventions, data storage, testing, CI/CD, etc.) based on requirement characteristics. - Memory-first research: Mandates ki-search project memory queries and asset reuse (expert-lookup / solution-lookup) before any code search, avoiding redundant investigation. - Parallel search acceleration: When two or more dimensions require code search, it dispatches parallel sub-agents via task-dispatch and merges results. - Use Case: Before designing a JWT authentication feature, run a survey to locate existing auth code, confirm API response conventions, and check error-handling patterns, producing a code-survey.md reference document. ## Quick Start Ask the agent to survey the existing codebase for the dimensions relevant to your current requirement and save the findings as code-survey.md in the requirement's reference directory.

Frequently Asked Questions about code-survey

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I research an existing codebase before designing a new feature?

Run a scoped code survey that selects only the dimensions relevant to your requirement, such as code paths, architecture, and API conventions. The results are written to a code-survey.md file in the requirement's reference directory for design use.

What dimensions does a codebase survey cover?

It covers 13 dimensions including code style, code paths, architecture, similar features, tech stack, data storage, error handling, security, logging, API conventions, testing, configuration, and CI/CD. Only dimensions matching the requirement's characteristics are investigated.

When can the code survey step be skipped?

It can be skipped or minimized when context maturity is high, the user already knows the project code, or project memory already contains sufficient architecture knowledge. Otherwise the survey is mandatory before design.

Does the survey search code directly or check project memory first?

Project memory via ki-search is always queried first, followed by mandatory asset reuse checks through expert-lookup and solution-lookup. Direct code search only happens for dimensions the first two steps cannot cover.

How does parallel code search work for multiple dimensions?

When two or more dimensions need code search, the survey dispatches parallel sub-agents via task-dispatch, one per dimension. Each sub-agent writes its findings to a dimension file, which the main agent merges into the final code-survey.md.