Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
cruldra avatar

Mapping Domain Vocabulary

  • 1 installs
  • Updated June 4, 2026
  • cruldra/skills

Helps with ai & agent building tasks.

About

mapping-domain-vocabulary is a Claude Code skill for ai & agent building. It helps developers move faster with AI-assisted coding.

  • mapping-domain-vocabulary
  • AI & Agent Building
  • AI-coding skill

Mapping Domain Vocabulary by the numbers

  • 1 all-time installs (skills.sh)
  • Ranked #14,102 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
  • Data as of Jul 7, 2026 (Skillselion catalog sync)
npx skills add https://github.com/cruldra/skills --skill mapping-domain-vocabulary

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1
Last updatedJune 4, 2026
Repositorycruldra/skills

What it does

Helps with ai & agent building tasks.

Files

SKILL.mdMarkdownGitHub ↗

Mapping Domain Vocabulary

Overview

The fastest index into complex cross-subsystem code is not the call chain — it is the domain vocabulary. Read the code as a dictionary: exhaustively extract every domain word (models, enums, fields, actions), give each a one-sentence role, and organize them into an ASCII tree grouped by subsystem → module → word.

Core principle: vocabulary before mechanism. Newcomers don't get stuck on flows; they get stuck on nouns. Once hold, entitlement, subminor are understood as words, the call chains read themselves.

When to Use

  • The user asks to "survey the code for X" / "help me quickly understand domain Y" (梳理/盘点)
  • The domain spans 2+ directories/services/packages (e.g. commerce + billing + sdk)
  • You need a panoramic grasp of a subdomain before design or refactoring

NOT for: single-file/single-module questions (just read it); questions about one specific flow (explain the call chain instead).

Process

1. Draw boundaries: split the domain into 2–4 subsystems by directory/service/package. 2. Fan out in parallel: dispatch one Explore subagent per subsystem (all in a single message) using the prompt template below, demanding exhaustive collection. 3. Merge into a tree: in the main session, dedupe and merge, then organize an ASCII tree by layer → module → word (format per the explaining-with-ascii-tree skill). Leaf = word ...... one-sentence role. 4. Dualities after the tree: cross-layer synonyms/dual terms (e.g. order↔entry), flow directions, and unit conventions go in 1–2 short paragraphs after the tree — never a paragraph per branch. 5. Archive: the output is ready to save as-is (Obsidian/docs).

Subagent Prompt Template

Explore <subsystem path> (<one-line description>), search breadth: very thorough.
Goal: collect all domain vocabulary related to <domain> (English words / identifiers).
1. Read the module's README/docs first (if any)
2. Sweep every submodule's models, schemas, service, routes
Return a list grouped by module, one term per line:
word | role (one sentence, in the conversation language) | path:line
Be exhaustive: model class names, every enum value, key concept fields,
task names, external integration concepts, exceptions. Err on the side of
too many. Return raw data — no polish needed.

Common Mistakes

MistakeFix
Writing a call-chain narrative (## 1/2/3 headers + paragraphs)Vocabulary is the index; mechanism goes in 1–2 paragraphs after the tree
Cherry-picking "highlights", skipping enum values/states/fieldsExhaustiveness is the whole point — err on the side of too many
Exploring sequentially in one threadFan out per subsystem in parallel — saves time and main-session context
Writing leaves as full paragraphsLeaf = word + one-sentence role; split anything longer than one line
Dropping code locationsMark paths at branch level, path:line on key terms
Same concept named differently per layer, left unstatedDual terms (order↔entry, minor↔subminor) must be listed explicitly

Example

The payment/billing domain of this repo (commerce + billing + sdk, three subsystems) was swept in one round with 2 parallel subagents; output archived as the Obsidian note「付费扣费领域词汇表」.

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.