Agent skill

Ubiquitous Language

by CodeAlive-AI in CodeAlive-AI/ai-driven-development

Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles.

MITAuto-check: notesDevelopment

Install Ubiquitous Language

skills CLI
$ npx skills add CodeAlive-AI/ai-driven-development --skill ubiquitous-language -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install CodeAlive-AI/ai-driven-development ubiquitous-language --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/CodeAlive-AI/ai-driven-development.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/ubiquitous-language .claude/skills/ubiquitous-language && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
ubiquitous-language
GitHub stars
155
Token cost
~7.4k tokens
SKILL.md length
3,224 words
Files
6 (incl. scripts, references)
Skills in repo
22
Repo updated
First seen
Licence
MIT

At a glance

Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles.

  • Works in 3 steps: If the user specified a path — use it → If THESAURUS.md already exists somewhere… → Default: docs/THESAURUS.md
  • The user asks to create thesaurus
  • SKILL.md covers Foundations, Core Principle, Thesaurus File and Lookup Protocol, plus 5 more sections
  • Runs Python scripts from its folder; calls rg, python3 and git

What it does

Ubiquitous Language is an agent skill from CodeAlive-AI/ai-driven-development. Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles. Use PROACTIVELY when naming anything: variables, functions, classes, modules, database fields, API endpoints, events, files, or directories. Also use when the user asks to "create thesaurus", "update glossary", "add term", "rename to match domain", "check naming consistency", "what should I call this", "domain language", "ubiquitous language", or "naming conventions". Ensures all names in the codebase are consistent…

Its SKILL.md is about 7.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files, including scripts and reference files (for example `README.md`, `references/generating-thesaurus.md` and `references/git-history-mining.md`).

It sits in Development, covering Domain-driven design. The repository describes itself as: Practices, protocols, and skills for AI-driven software development. Skills and safety hooks for Claude Code, Codex, OpenCode, Cursor, Antigravity, and any agent supporting the… The licence is MIT.

When your agent uses it

  • The user asks to create thesaurus
  • Update glossary
  • Rename to match domain
  • Check naming consistency

Example prompts

  • “create thesaurus”
  • “update glossary”
  • “add term”
  • “/ubiquitous-language”

Requirements

  • Python 3
  • Pre-approved tools (allowed-tools): Read, Write, Edit, Grep, Glob, Bash, Agent, AskUserQuestion

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. If the user specified a path — use it
  2. If THESAURUS.md already exists somewhere in the repo — use that location
  3. Default: docs/THESAURUS.md

What it can do on your machine

Read from SKILL.md and the folder at commit 25b7b1d. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Write
    • Edit
    • Grep
    • Glob
    • Bash
    • Agent
    • AskUserQuestion

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • rg
    • python3
    • git

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Ubiquitous Language loads about 7.4k tokens when it runs, and up to ~23k if it reads all its reference files. Until then it costs about 199 tokens; SKILL.md has 3,224 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~199
When it runs · the whole SKILL.md, loaded when a task matches
~7.4k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~23k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Read, Write, Edit, Grep, Glob, Bash, Agent, AskUserQuestion

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from CodeAlive-AI/ai-driven-development at commit 25b7b1d, republished under its MIT licence (© CodeAlive-AI). 3,224 words, ~7,374 tokens.

Download SKILL.mdSave it as .claude/skills/ubiquitous-language/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.
name
ubiquitous-language
description
Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles. Use PROACTIVELY when naming anything: variables, functions, classes, modules, database fields, API endpoints, events, files, or directories. Also use when the user asks to "create thesaurus", "update glossary", "add term", "rename to match domain", "check naming consistency", "what should I call this", "domain language", "ubiquitous language", or "naming conventions". Ensures all names in the codebase are consistent, descriptive, and aligned with the shared domain vocabulary. Also mines git history to resolve naming ambiguities — when a name was born, which name replaced which, which spelling is dying. Not for general code style or linting — only for domain term consistency.
allowed-tools
Read, Write, Edit, Grep, Glob, Bash, Agent, AskUserQuestion
metadata.version
2.1.0
metadata.thesaurus-format
2.0

Ubiquitous Language: Project Thesaurus Manager

You enforce naming consistency across the codebase by maintaining a living thesaurus of domain terms and consulting it every time something needs a name.

Four modes:

Foundations

This skill combines two bodies of knowledge:

  • Domain-Driven Design (DDD) by Eric Evans — ubiquitous language, bounded contexts, aggregate naming
  • First Principles Framework (FPF) — a transdisciplinary "operating system for thought" that provides formal tools for semantic precision: bounded contexts as declared semantic frames, polysemy unpacking, lexical firewalls, cross-context bridges with loss notes, term continuity relations, and anti-explosion naming control. References like "FPF F.5" or "FPF A.1.1" point to specific sections of the FPF specification.

Core Principle

"A project should use a single, shared vocabulary. Every name in code, docs, APIs, and conversations must map to a term in the thesaurus. If a concept isn't in the thesaurus — add it before naming anything."

— Domain-Driven Design, Eric Evans

The codebase is primary evidence, not automatic authority. Use code to discover which terms are currently in circulation. Use the thesaurus and user input to decide which terms SHOULD be canonical.

  • For what exists today — derive from code (classes, DB schemas, API routes, events)
  • For what should become the standard — ask the user/domain expert
  • If code contradicts the approved thesaurus — the thesaurus wins for new code
  • If the user says "fix legacy naming" — the user's directive overrides the codebase; map existing code names to ## Legacy lines and use the user's terms as canonical

Tacit knowledge: For areas not yet implemented, the most important domain knowledge exists only in experts' heads, not in any artifact.

Thesaurus File

Locating the thesaurus:

  1. If the user specified a path — use it
  2. If THESAURUS.md already exists somewhere in the repo — use that location
  3. Default: docs/THESAURUS.md

Single source of truth for domain vocabulary.

Versioning

The thesaurus declares its format version and the skill that maintains it in YAML frontmatter — machine-readable, outside the body, still one grep away:

markdown
---
thesaurus-format: "2.0"
skill: ubiquitous-language
---

# Project Thesaurus

Quote the version — unquoted 2.10 is the YAML float 2.1.

  • rg -n '^thesaurus-format:' THESAURUS.md → the version in one hit. No key = 1.0 (the pre-index prose layout shipped before skill 2.0) — unless the file already has - **Term** `Id` kind: Index lines: that is an unstamped 2.0 file from plugin 9.2.0; just add the stamp.
  • Format major = skill major (metadata.version in this file's frontmatter). Skill minor/patch releases never change the format.
  • Read any format ≤ your own; write only the current one. A major gap means migration first (see generating-thesaurus.md); a minor gap means new optional tokens — older readers keep working, you may add the tokens as you touch lines.
  • A thesaurus with a format newer than yours: read it, don't rewrite it — tell the user to update the skill.
  • Only the format is versioned in the file. The skill's own version is not recorded there — it would go stale on every edit and git log already answers "who wrote this". skill: is a pointer, so an agent without the skill knows what to install.
FormatLayoutSkill
1.0### Term entries with Synonyms to AVOID; ## Legacy Terms entries; ## Forbidden Lexicon table≤ 1.x (plugin ≤ 9.1.1)
2.0grep-first: ## Index lines with kind:/ctx:/avoid:, use: Forbidden lines, → Legacy lines, SKOS bridges2.x
Layout: grep-first

The file is designed so that one rg/grep for any name answers "what do I do with this name?" without reading the surrounding text. Five sections, fixed order:

SectionShapeOne line answers
## Indexone line per concept"Is there a term for this? Which name is canonical? What is banned?"
## Terms### Term entries"What exactly does it mean / not mean / relate to?"
## Forbiddenone line per word"Is this word banned from domain names?"
## Legacyone line per old name"This old name is in the code — what replaced it?"
## Unresolved### Term — problem entries"Is this name an open question?"

Registry invariant: every name known to the project appears in exactly one registry line — an Index line (as Term, Identifier, or avoid), a Forbidden line, a Legacy line, or an Unresolved header. Each kind of registry line has its own shape, so the shape of the hit tells you its status and the line itself tells you the canonical name. No -B/-A context needed.

Registry lines are bullet lines with labelled tokens, not Markdown tables: tables need | (an alternation in rg), match by column position, and get re-padded by formatters. Tokens (kind:, ctx:, avoid:, use:, in:, →) are position-free, formatter-proof, and need no escaping.

markdown
# Project Thesaurus

## Index

- **Order** `Order` kind:aggregate avoid: `Purchase`, `Transaction`, `Buy`
- **Order Line Item** `OrderLineItem` kind:entity avoid: `LineItem`, `OrderItem`, `Item`
- **Order Placed** `OrderPlaced` kind:event avoid: `OrderCreated`, `NewOrder`

## Terms

### Order
- **Definition**: A customer's confirmed request to buy one or more products at agreed prices.
- **NOT**: A payment (that's `Payment`), a shipment, or a draft cart (that's `Cart`).
- **Related**: Order Line Item, Order Placed, Cart

## Forbidden

- `Manager` use: `OrderFulfillment` — hides responsibility; name the activity

## Legacy

- `UserManager` → `Customer` + `CustomerRegistration` in: `src/legacy/` — split in v3

## Unresolved

### Account — one word, two concepts (billing vs auth)
- **Found in**: `billing/Account.ts` (balance), `auth/Account.ts` (login)
- **Question**: Two bounded contexts, or one of them a naming mistake?
- **Impact**: 18 files
- **Options**: `BillingAccount` + `UserAccount`; or contexts Billing / Identity
Index line
- **<Term>** `<Identifier>` kind:<kind> [ctx:<Context>] [avoid: `<name>`, `<name>`]
FieldContentRules
TermHuman name, as domain experts say it, in **bold**May be multi-word or non-English. Also the ### header text of the entry
IdentifierPascalCase code form, in backticksThe thing you grep in code. All other casings derive from it mechanically (see Casing)
kind:One of aggregate entity value event command query service role process state policy conceptPicks the naming rule below. concept when nothing fits
ctx:Bounded context nameOnly present once contexts are confirmed (see generating-thesaurus.md); the header then becomes ### Term (Context)
avoid:Banned synonyms and abbreviations, each in backticks, comma-separatedLast on the line because it is the only variable-length field. Omit when empty
  • One line per concept. All names for that concept live on that line — this is what makes reverse lookup (rg Purchase → "use Order") a single hit.
  • Backticks around every identifier-like name. rg '`Order`' is an exact match; rg Order would also hit OrderLineItem and Reorder. rg -F '**Order**' is the exact Term.
  • Sorted alphabetically by Term. Entries under ## Terms follow the same order.
  • avoid lists live only here. Entries do not repeat them — one place, no drift.
Forbidden and Legacy lines
- `<Word>` use: `<Identifier>`[, `<Identifier>`] — <why>
- `<OldName>` → `<Identifier>`[ + `<Identifier>`] in: <files/modules> — <note>
- `<OldName>` → — (see `<X>`, `<Y>`) in: <files> — retired

use: always points at an Index Identifier. → is the legacy marker: A + B means the old name was split; → — means retired with no single successor.

Entry
markdown
### [Term]
- **Definition**: What this concept means in the business domain — one sentence
- **NOT**: What this term does NOT mean; name the neighbouring term it is confused with
- **Related**: Other Index Terms this connects to, written exactly as in the Term field

Optional lines when they carry real information: **Broader**, **Narrower**, **Part of**, **Has parts**, **Example**. Write them on one side only — rg gives the inverse for free, mirrored copies only drift. Minimal viable entry is one line: - **Definition**: … — the Index line already holds the rest.

Anchor grammar (so lookups are one regex): the header is exactly ### <Term> or ### <Term> (<Context>) — nothing else. Tags, status, and context prefixes belong in the Index line, not in the header. Find an entry with rg -n '^### Order( \(|$)' — \b alone is not enough, it would also match ### Order Line Item.

The thesaurus captures concepts, not behavior. It's strong at nouns (entity names, roles, process names) but won't replace behavioral specs for business rules. Don't try to turn the thesaurus into a specification — keep entries short. If a concept has a critical invariant, note it briefly in the definition, not as a separate section.

Non-English domains: If the business domain operates in a non-English language, the Term uses the original language — the thesaurus should reflect how domain experts actually speak. The Identifier carries the code form: - **Счёт-фактура** `Invoice` kind:entity. This is exactly why both fields exist.

Lookup Protocol

Look up before inventing. This is the single most important step. Most naming tasks don't need a new term — the right name is already there.

  1. Locate the thesaurus (see above). If absent, tell the user and offer generation. Check rg -n '^thesaurus-format:' — no key or 1.x means the old layout: the protocol below still works by plain text search, but offer migration once, up front.

  2. Read the Index if it has ≤ ~60 lines — it is the entire vocabulary at one line per concept, cheaper than any search. For larger files, search instead.

  3. Search every candidate name you are considering, plus whatever the surrounding code already calls the thing. Use rg (the Grep tool) or grep -E — same patterns:

    bash
    rg -n -i 'invoice' docs/THESAURUS.md             # any role, any section
    rg -n '`OrderLineItem`' docs/THESAURUS.md        # exact identifier as seen in code
    rg -n -F '**Order**' docs/THESAURUS.md           # exact Term (not "Order Line Item")
    rg -n 'avoid:.*`Purchase`' docs/THESAURUS.md     # is this word a banned synonym?
    rg -n 'kind:event' docs/THESAURUS.md             # all terms of one kind
    rg -n 'ctx:Billing' docs/THESAURUS.md            # everything one context owns
    rg -n '`Basket` →' docs/THESAURUS.md             # legacy name and its replacement
    rg -n -A4 '^### Invoice( \(|$)' docs/THESAURUS.md  # the entry itself

    The only trap: * and | are regex metacharacters — use -F for **Term**, and never search for table pipes (there are none).

  4. Act on the shape of the line you hit:

    Hit line looks likeMeaningDo
    - **X** `X` kind:… — your word is the Term or IdentifierConcept existsUse the Identifier exactly. Stop
    - **X** … avoid: … `your word` You were about to use a banned synonymUse that line's Identifier instead
    - `word` use: …Word is banned from domain namesPick the Identifier after use:
    - `word` → …Old name still in codeUse the replacement after → for new code; don't spread the legacy name
    ### word — … under ## UnresolvedOpen naming questionDon't decide silently — surface it, ask the user, or offer history mining (see below)
    ### header / entry text onlyRelated conceptRead the entry; it may inform composition
    No hitNew conceptGo to "If the concept is new"
  5. Check the bounded context if Index lines carry ctx: — the same word may be canonical in one context and banned in another.

ALWAYS run this before naming: classes, interfaces, types, enums, aggregates, entities, value objects, functions, methods, commands, queries, domain events, variables, constants, fields, parameters, DB tables/columns/collections, API endpoints and response fields, files, directories, modules, packages, feature flags, config keys, environment variables, and commit messages or PR titles that reference domain concepts.

If the Concept Is New

Before minting a new term, try four levers (from FPF F.14 "Name less, express more"):

  1. Reuse — does an existing term already cover this? Maybe the concept is a variant, not a new thing
  2. Compose — can you combine existing terms? OrderLineItem reuses Order + LineItem
  3. Qualify — is this the same concept in a different state/window? Don't create NightOperator — use Operator with a time qualifier
  4. Ask — if still unclear: "I need to name [concept]. The thesaurus doesn't have a term for this. What does the domain call it?" If the user doesn't have an answer either — that's a white spot, not a dead end. Building a ubiquitous language is co-creation, not extraction. Add it to ## Unresolved with a [WHITE-SPOT] tag in the header. Don't force a name for an undefined concept.

Only after all four fail, mint a new term:

  1. Name what the invariants make true (FPF F.5) — don't name aspirationally. If the code doesn't enforce "Premium", don't call it PremiumCustomer
  2. Use minimal generality — choose the narrowest name whose rules you actually enforce. Don't upgrade Task to Activity to sound universal
  3. Keep it to 1-3 words — no rhetorical adjectives ("robust", "optimal", "advanced")
  4. Add it to the thesaurus: an Index line (Term, Identifier, kind:, avoid: — omit avoid: when empty) and a ### Term entry with at least a Definition. Keep both sorted
  5. Then use the term in code

If You Find an Inconsistency

When existing code uses a term that contradicts the thesaurus:

  • Flag it: "Found fetchPurchases() but the Index line says Order, with Purchase under avoid:"
  • Suggest a rename if scope is small
  • For large-scale renames, note as tech debt and ask user how to proceed
Show full SKILL.md (1,409 more words)Show less

If the Ambiguity Won't Resolve Itself

When a name lands in ## Unresolved — two spellings for one concept, one spelling for two concepts, no obvious winner — the current tree can't settle it, but the repository's history often can: which identifier was born first, which commit removed one while adding the other, which one is dying.

Offer it, don't run it silently. Whenever you present ## Unresolved items — after generation, after an audit, or when a naming question hits one — offer once:

"I can mine the git history for these — when each name was born, which replaced which, which is growing vs dying. Temporary index outside the repo, deleted afterwards. Want me to try that before you answer them by hand?"

Skip the offer if there is no .git, history is shorter than ~50 commits or squashed from an import, or the items are [WHITE-SPOT] tags (an unnamed concept leaves no trace).

If the user accepts, read references/git-history-mining.md and follow it. The short version:

bash
S=<skill-dir>/scripts/git_term_index.py
python3 $S build --repo-dir . --content   # throwaway SQLite index in $TMPDIR, never in the repo
python3 $S query Account Customer         # birth, dormancy, trajectory, renames, messages
python3 $S pair User Customer             # competing names: birth order + swap commits
python3 $S contexts Account               # where the name lives — the polysemy check
python3 $S search 'rename account'        # BM25 search over commit messages
python3 $S clean                          # delete the index when done

The index covers the whole diff history, so "born" means born. Use --pathspec src/ on large repos — it cuts build time and sharpens the signal at once.

pair ends in a labelled verdict — RENAME (strong / probable / possible), DRIFT, NOT A RENAME, COEXISTENCE — with the direction inferred from evidence, not argument order. Report the label as given; do not upgrade it. "RENAME — strong" means commits exchange the names and a subject announces it; everything weaker needs git show first.

For polysemy (one word, two meanings, two modules) use the path split: pair prints files: A in N, B in M, both in K, and contexts <name> gives the directory breakdown for a single word. both in 0 — no file ever contained both — is real evidence for two bounded contexts; shared files mean synonym drift, which history cannot settle. Only claim a path split when the tool printed one. File-only renames (the file moved, the identifier did not) appear in query's file-renames section, not in pair — run both.

History is evidence, not authority. It ranks candidates and cites commits; the user decides. Report proposals in one batch with confidence levels, apply only what is approved, and record the commit behind each applied decision (— renamed in a41f2c9 on the Legacy line, or a - **History**: line on the entry).

Naming Rules by DDD Construct

The Index Kind column selects the rule.

Aggregates & Aggregate Roots (aggregate)

Use the business domain term. Singular. No technical suffixes.

GOOD: Order, Invoice, UserAccount, ShoppingCart
BAD:  OrderAggregate, OrderRoot, OrderAggregateImpl, OrderEntity
Entities (entity)

Singular noun from the domain. Something with identity.

GOOD: OrderLineItem, PaymentTransaction, Customer
BAD:  OrderLineItemEntity, OrderLineItemImpl, OrderLineItemObj
Value Objects (value)

Singular noun describing an immutable concept. Describes what it is, not what it does.

GOOD: Money, Email, PhoneNumber, Address, DateRange
BAD:  MoneyValue, EmailValidator, PriceInfo, AmountData
Domain Events (event)

Past tense verb + noun. Something that happened.

GOOD: OrderPlaced, PaymentCaptured, InvoiceSent, InventoryReserved
BAD:  OrderEvent, OnOrderPlaced, CreateOrder (that's a command)
Commands (command)

Imperative verb + noun. An action requested.

GOOD: CreateOrder, CancelInvoice, ProcessRefund, ReserveInventory
BAD:  OrderCreated (that's an event), NewOrder, OrderCommand
Queries (query)

Question or retrieval. Verb + object or descriptive name.

GOOD: GetOrderById, FindInvoicesByCustomer, ListPendingOrders
BAD:  RetrieveOrderData, OrderQuery, GetterForOrder
Domain Services (service)

Named after business activities the domain expert recognizes.

GOOD: InvoiceCalculator, OrderFulfillment, NotificationSender
BAD:  OrderManager, GenericService, HelperService
Repositories

Repository suffix is acceptable — it's an infrastructure pattern. Repositories are not thesaurus terms; they take the name of the aggregate they store.

GOOD: OrderRepository, InvoiceRepository, CustomerRepository
BAD:  OrderStorage, OrderPersistence, OrderFinder, OrderDao
Methods on Aggregates

Commands (change state): Imperative verb, no "Get" prefix.

GOOD: order.Cancel(), order.AddLineItem(product, quantity), order.Recalculate()
BAD:  order.CancelOrderMethod(), order.GetCancelled(), order.DoCancelOrder()

Queries (read-only): Start with Get, Is, Has, Can, or a domain verb.

GOOD: order.GetTotal(), order.IsExpired(), order.CanBeShipped()
BAD:  order.FetchInfo(), order.CheckData()

Naming Anti-Patterns to Detect and Flag

Lexical Firewall: the ## Forbidden section

The domain layer must be protected from transient jargon, vague terms, and implementation details. The thesaurus's ## Forbidden section lists words that MUST NOT appear in domain names and must always be replaced with a specific domain term.

Weasel Words (never use in domain layer)
Weasel WordProblemFix
InfoMeaningless suffixRemove it: UserInfo -> User
DataSays nothing about the conceptUse domain term: OrderData -> Order
ManagerVague, hides responsibilitySplit by actual responsibility
HandlerGeneric, unclear intentName after what it handles
ServiceOverused catch-allUse specific domain activity name
BaseTechnical distractionRemove, use composition
ItemToo genericUse domain term: Item -> OrderLineItem, Product
Util / HelperIndicates bad designMove logic to domain objects
Object / ObjNever appropriateRemove suffix
Record / ModelDatabase concept leaking into domainUse domain term
Config / SettingsGeneric container hiding a conceptConfig -> LoanProduct, Settings -> NotificationPreferences
Technical Jargon in Domain Layer

Domain code must be free of implementation details:

BAD:  MongoOrder, SqlUserRepository, HttpOrderService, OrderDto, OrderEntity
GOOD: Order, OrderRepository (interface), PaymentGateway, Order (just Order)

Technical prefixes/suffixes belong ONLY in the infrastructure layer, and even there the domain role should lead:

INFRASTRUCTURE LAYER (OK): MongoOrderRepository, RedisSessionCache, HttpPaymentClient
DOMAIN LAYER (NEVER):      MongoOrder, RedisSession, HttpPayment
NAME THE ROLE, NOT THE TECH: SessionStore not RedisCache, EventPublisher not KafkaProducer

Framework caveat: In frameworks that intentionally blend domain and persistence (Active Record pattern, ORM-centric frameworks), the model IS the domain entity. Keep the domain noun clean and let framework coupling live in inheritance, annotations, or metadata — not in the class name. Flag technical jargon only when it becomes part of the business-facing name or leaks outside its boundary.

Synonym Drift

Same concept called different things in different parts of code:

PROBLEM: "Customer" in auth, "User" in API, "Account" in billing — all mean the same thing
FIX:     Pick ONE canonical term per bounded context. Put the others in that line's `avoid:` list.
Abbreviation Boundary

Ban abbreviations in durable, domain-bearing names: types, exported functions, modules, API fields, DB columns, events, config keys.

Allow conventional short-lived local identifiers when meaning is obvious in scope: i, j, ctx, req, res, err, tx, db, e for events.

Allow industry-standard acronyms when they are the dominant term: SKU, VAT, URL, ID, OAuth. Do NOT force unnatural expansions if experts use the acronym.

PROBLEM: usr, user, account, acct — competing abbreviations for the same durable concept
FIX:     Pick ONE canonical form for domain-bearing names. Short-lived locals are exempt.
Translation Chain ("Telephone Game")

When different artifacts use different terms for the same concept across the knowledge chain, information is lost at each translation:

SMELL: Domain expert says "Campaign" → PM writes "Promotion" in spec →
       Dev codes `marketing_push` → QA tests "advertising effort"
FIX:   Same term everywhere: expert, PM, dev, QA all say and write "Campaign"

This is worse than synonym drift because each translation also loses nuance and business rules. How to detect: compare terms in requirements/specs/tickets against code names. If they don't match, the ubiquitous language has a translation gap — adopt the domain expert's term everywhere.

Casing Conventions

The Index Identifier is PascalCase. Every other form is derived from it mechanically using the project's conventions — never re-worded:

ContextConventionExample (Identifier: OrderLineItem)
Class/TypePascalCaseOrderLineItem
Function/MethodProject conventionaddOrderLineItem / add_order_line_item
VariableProject conventionorderLineItem / order_line_item
ConstantUPPER_SNAKEMAX_ORDER_LINE_ITEMS
Database tableProject conventionorder_line_items
API endpointkebab-case or convention/orders/{id}/line-items
Event/MessagePascalCase with past-tense verbOrderLineItemAdded
File/DirectoryProject conventionorder_line_item.py, OrderLineItem.cs

Key rules:

  • Use the EXACT Identifier — don't abbreviate (ord), don't expand (orderObject), don't synonym (purchase)
  • Compound names combine Identifiers: OrderLineItem, not PurchaseLineItem
  • Technical suffixes for infrastructure roles are fine: OrderRepository, OrderDTO (in infra layer only)
  • Multi-word Identifiers keep all words in every casing: ProcessingStage → processing_stage, never proc_stage
  • Variables and parameters use full descriptive names: totalAmount not amt, customerEmail not cEmail

Updating the Thesaurus

When changing terms, use the least strong relation that tells the truth (from FPF F.13):

OperationWhenEffect on thesaurus
AddNew conceptIndex line + entry. Minimum: Identifier, kind:, Definition
RenameWording improved, sense unchangedChange Term/Identifier in the Index line and header; old Identifier → ## Legacy line `Old` → `New`; grep codebase, suggest renames
SplitOne term covered two sensesOld line removed; two new lines + entries; old Identifier → ## Legacy line `Old` → `A` + `B`; disambiguation in each NOT
MergeTwo terms are really one senseKeep one line; the other Identifier moves into its avoid: list; entries merged
RetireTerm was misleading, no single successor## Legacy line `Old` → — (see `X`, `Y`)
DeprecateConcept being phased out## Legacy line with → replacement and in: locations

Key test: Can you point to the same concept before and after the change?

  • Yes, same concept, better wording → Rename (keep as legacy alias for reading old code)
  • No, the concept actually changed → Split or Merge (not a rename)

Alias parsimony: keep at most 1 legacy alias per term — the one readers will most likely encounter in old code. Registry invariant still holds after every edit: a name is under avoid: or in Legacy, never both.

Old layout? No thesaurus-format frontmatter key (or 1.x) means the pre-index prose layout — see "Migrating an Existing Thesaurus" in references/generating-thesaurus.md.

Quick Checklist Before Naming Anything

  1. Did I grep the thesaurus for this name and its synonyms? Where did the hit land?
  2. Would a domain expert recognize this name?
  3. Does it contain a weasel word (Manager, Service, Handler, Info, Data, Item, Base, Util)?
  4. Is it too generic (could mean multiple things in different contexts)?
  5. Does it reveal infrastructure details (Mongo, Sql, Http, Dto, Entity, Model)?
  6. Is it consistent with other uses of this term across the codebase?
  7. Am I using the EXACT Identifier from the Index, or a synonym from its avoid: list?
  8. Am I in the right bounded context for this term?
  9. Does the name match its Kind — past tense for event, imperative for command?
  10. Can I explain what this name represents in one sentence using domain language?
  11. If the concept is new — did I add the Index line and the entry before using it?

If any answer raises a concern — stop and fix before proceeding.

© CodeAlive-AI, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (scripts, references) in skills/ubiquitous-language of CodeAlive-AI/ai-driven-development.

  • SKILL.md
  • README.md
  • references/generating-thesaurus.md
  • references/git-history-mining.md
  • references/naming-audit.md
  • scripts/git_term_index.py

Open the folder on GitHubat commit 25b7b1d

Compare with similar skills

Ubiquitous Language next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Ubiquitous Language compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Ubiquitous Language this skillCodeAlive-AI/ai-driven-development155—~7.4kAutomated safety check: NotesMIT
Domain Modelingfossasia/eventyay-interpretation1.6k29 repos~821Automated safety check: PassApache-2.0
Architecture Governancezai-org/ZCode7.5k—~1.2kAutomated safety check: PassApache-2.0
Evolutionary Modular Architecturetech-leads-club/agent-skills7k—~3.7kAutomated safety check: PassCC-BY-4.0
Domain Modelingbrim-borium/spotify_sdk1665 repos~806Automated safety check: PassApache-2.0
Domain Modeling and Glossarywindmill-labs/windmill18k—~622Automated safety check: PassCustom licence

Similar skills

  • Domain Modeling

    fossasia/eventyay-interpretation

    Build and sharpen a project's domain model. An agent skill from fossasia/eventyay-interpretation.

    1.6k GitHub starsUsed in 29 repos~821 tokens
    DevelopmentAuto-check passed
  • Apply the repository's architecture policy to code changes by generating a bounded context package, checking module and layer boundaries, and reporting baseline-aware violations.

    7.5k GitHub stars~1.2k tokensUpdated 8 days ago
    DevelopmentAuto-check passed
  • Evolutionary Modular Architecture

    tech-leads-club/agent-skills

    Guides design of modular-monolith platforms with DDD, flat-by-aggregate modules, anti-corruption layers, outbox events and resilience, plus an architecture document with SVG diagrams.

    7k GitHub stars~3.7k tokensUpdated 17 days ago
    DevelopmentAuto-check passed
  • Domain Modeling

    brim-borium/spotify_sdk

    Build and sharpen a project's domain model. An agent skill from brim-borium/spotify_sdk.

    166 GitHub starsUsed in 5 repos~806 tokens
    DevelopmentAuto-check passed
  • Domain Modeling and Glossary

    windmill-labs/windmill

    Actively challenges vague or conflicting terminology as you design, and keeps a living domain glossary file up to date in real time.

    18k GitHub stars~622 tokensUpdated yesterday
    DevelopmentAuto-check passed
  • Ddd

    swamp-club/swamp

    Domain Driven Design guidance for TypeScript/Deno codebases.

    642 GitHub stars~1.4k tokensUpdated yesterday
    DevelopmentAuto-check passed

More from CodeAlive-AI/ai-driven-development

All 22 skills in this repo
  • Investigating Repository History

    CodeAlive-AI/ai-driven-development

    Investigate GitHub repository history before risky code changes using git blame/log, GitHub PRs, review comments, squash/rebase/cherry-pick/rename heuristics, and cited evidence.

    155 GitHub stars~2k tokensUpdated yesterday
    Auto-check passed
  • Plugins Management

    CodeAlive-AI/ai-driven-development

    Create, publish, delete, and submit plugins for coding agents (Claude Code, OpenCode, Devin CLI/Desktop).

    155 GitHub stars~3.1k tokensUpdated yesterday
    Auto-check: notes
  • Semantic Scholar Deep

    CodeAlive-AI/ai-driven-development

    Deep research over the Semantic Scholar Graph API. An agent skill from CodeAlive-AI/ai-driven-development.

    155 GitHub starsUsed in 1 repo~2.2k tokens
    Auto-check passed
  • Windows QA Engineer

    CodeAlive-AI/ai-driven-development

    A skill your agent uses when testing Windows 11 desktop apps (WinForms/WPF/UWP) via UFO UIA/Win32 automation MCP.

    155 GitHub stars~1.4k tokensUpdated yesterday
    Auto-check passed
  • Agentic Readiness

    CodeAlive-AI/ai-driven-development

    Audit and improve repositories for reliable agentic work across Codex and Codex App, Claude Code, and OpenCode.

    155 GitHub stars~1.2k tokensUpdated yesterday
    Auto-check passed
  • Hooks Management

    CodeAlive-AI/ai-driven-development

    Manage hooks and automation for coding agents (Claude Code, Codex CLI, OpenCode, Devin CLI/Desktop).

    155 GitHub stars~4.8k tokensUpdated yesterday
    Auto-check: notes

Categories

Questions about Ubiquitous Language

What does Ubiquitous Language do?

Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles. Ubiquitous Language is an agent skill from CodeAlive-AI/ai-driven-development. Maintain a project thesaurus (domain glossary) following DDD ubiquitous language principles.

When should I use Ubiquitous Language?

Ubiquitous Language fits situations like: the user asks to create thesaurus; update glossary; rename to match domain; check naming consistency.

How do I install Ubiquitous Language in Claude Code?

Run `npx skills add CodeAlive-AI/ai-driven-development --skill ubiquitous-language -a claude-code`. Or copy the skill folder (skills/ubiquitous-language in CodeAlive-AI/ai-driven-development) into .claude/skills/ubiquitous-language in your project. Claude Code loads it when a task matches its description.

How do I install Ubiquitous Language in Codex?

Run `npx skills add CodeAlive-AI/ai-driven-development --skill ubiquitous-language -a codex`. Or copy the skill folder (skills/ubiquitous-language in CodeAlive-AI/ai-driven-development) into .agents/skills/ubiquitous-language in your project. Codex loads it when a task matches its description.

Can I use Ubiquitous Language in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add CodeAlive-AI/ai-driven-development --skill ubiquitous-language -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ubiquitous-language, .gemini/skills/ubiquitous-language, .github/skills/ubiquitous-language and .opencode/skills/ubiquitous-language in your project.

What does Ubiquitous Language need to run?

Going by SKILL.md and its folder, Ubiquitous Language needs Python for the scripts in its folder and the command-line tools its instructions call (rg, python3 and git). Our summary lists: Python 3. Its frontmatter pre-approves these tools: Read, Write, Edit, Grep, Glob, Bash, Agent, AskUserQuestion.

Does Ubiquitous Language access the network?

SKILL.md names 1 domain. As links in the text: github.com. This is read from the text; nothing was executed.

Is Ubiquitous Language safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Ubiquitous Language use?

Ubiquitous Language is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Ubiquitous Language use?

About 7.4k tokens (SKILL.md is roughly 29k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 16k tokens, read only when the agent opens those files.

What are the alternatives to Ubiquitous Language?

Skills that share tags, products or a category with Ubiquitous Language: Domain Modeling (fossasia/eventyay-interpretation, 1.6k stars), Architecture Governance (zai-org/ZCode, 7.5k stars), Evolutionary Modular Architecture (tech-leads-club/agent-skills, 7k stars) and Domain Modeling (brim-borium/spotify_sdk, 166 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Ubiquitous Language?

CodeAlive-AI (a GitHub organization) maintains it in CodeAlive-AI/ai-driven-development, which has 155 GitHub stars. The repository holds 22 skills in this directory. The repository was last updated on October 6, 2026.

Source: CodeAlive-AI/ai-driven-development on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.