Dto Creator

Preflight: Spring MCP

This skill is part of the Spring Agent Toolkit and is designed to work with the Spring MCP server (provided by the Amplicode IntelliJ plugin). Before doing anything else, check your tool list for any Spring MCP tool — they are exposed under the amplicode MCP server (e.g. get_project_summary, list_module_dependencies, get_entity_details); harnesses that flatten MCP tools into the tool list use the mcp__amplicode__ prefix on the same names.

• If at least one Amplicode tool is available — MCP is connected. Proceed with the skill below.
• If none are available — stop and invoke the amplicode-install skill (bundled with the Spring Agent Toolkit). It installs the Amplicode plugin and walks the user through the «Настроить Spring Agent» welcome-screen button + MCP-client restart. After it completes, the MCP tools become available — resume this skill.
• If amplicode-install is not registered in your skill list, tell the user (in their language): "This skill needs the Amplicode IntelliJ plugin and its MCP server. Install it from https://amplicode.ru/marketplace into IntelliJ IDEA Ultimate/Community or GigaIDE, open any project, click «Настроить Spring Agent» on the Amplicode welcome screen, then restart your MCP client."

DTO Creator

Creates a DTO class (Java class, Java record, Java+Lombok, or Kotlin data class) for an entity with selected attributes, constructors, getters/setters, equals/hashCode, toString, and optional features.

CRITICAL: Code ONLY from examples/ files. If no matching example -- STOP and ask user. CRITICAL: For questions with a fixed set of choices, prefer AskUserQuestion > its analogue > plain text list. Plain numbered text lists are the last resort when no interactive tool is available. CRITICAL: Read the conversation context BEFORE running Step 1. Half the questions in Steps 2–7 may already be answered by the user's prompt and prior turns. Re-asking what was already said is the #1 reason this skill feels slow. Step 7 (mapper) is automatic when conversion is needed. If from context it is clear that the DTO will be used in code that converts entities to/from DTOs (controller, service, endpoint replacement, etc.), the skill MUST delegate to mapper-creator — never write manual mapping code inline. The mapper-creator skill decides the implementation (MapStruct, Custom, adding dependencies) — this skill just delegates.

Defaults

Smart defaults: If user says "use defaults", "all defaults", "default settings", or similar -- skip ALL questions where "Always ask?" = NO. Only ask mandatory questions.

Smart answer recognition: When user provides a value instead of choosing from a numbered list, accept it directly. Examples:

• Question "Which entity?" -> user answers "Order" -> this IS the entity, don't re-ask
• Question "Variant?" -> user answers "record" -> this IS Java record, don't show options
• If user provides multiple answers in one message -> accept all, skip answered questions
• NEVER ask a question that the user already answered (even implicitly)

Batch questions: Group closely related questions into a single AskUserQuestion call (up to 4 questions per call) when they:

• Belong to the same logical section (e.g. both are Java method generation settings)
• Don't depend on each other's answers
• Have obvious defaults that the user can skip

Rules:

• Maximum 3-4 questions per AskUserQuestion call
• Mark the recommended option with (Recommended) and place it first
• Never batch questions from DIFFERENT decision branches
• The primary branching question (entity selection, variant) is always asked ALONE
• Prefer AskUserQuestion for choices; fall back to plain text lists only if the tool is unavailable

Decision-making principle — context first, then ask

Before asking the user any question, attempt to derive the answer from the context already gathered: project summary, module dependencies, entity details, existing files in the package, prior turns of this conversation, and the user's original prompt. Only ask when the context yields no clear default or when the choice is genuinely user-specific (e.g. which entity, which fields).

Hierarchy of decisions:

1. Context is unambiguous → decide silently, do NOT ask. Examples: language and JDK from get_project_summary; Lombok / Jackson / Hibernate Validator from list_module_dependencies; package from the entity FQN; className from {Entity}Dto; back-reference @ManyToOne filtering; auto-selection of sub-entity scalars; record vs. plain class when JDK ≥ 16 and no Lombok and the user did not request otherwise. Exception: subDtoType is NEVER decided silently — it always requires at minimum principle 2 (one-line confirmation), unless the USER explicitly stated the shape in their own message (not in ARGUMENTS).

2. Context gives a strong signal → state the decision + alternatives in one line, let the user override or stay silent. Format:

   Will create `ScheduleTemplateDto` as a Java record with nested `SlotDto` (record).
   Alternatives: plain class, Lombok, separate file for SlotDto. OK?
   

   The user can answer "ok" / "yes" / silence → accept; or name an alternative → switch. This is not the same as the numbered question format — it is a single confirmation line.

3. Context yields no clear default → ask with AskUserQuestion (preferred) or its analogue, with the recommended option first. When AskUserQuestion is available, use it with preview fields so the user sees the concrete code shape for each option. Mark the recommended option with (Recommended) in its label and place it first. If no interactive choice tool is available, fall back to a plain text list. Never ask iteratively ("which variant?" → user picks → "which fields?" → …) when one batched call would do.

4. Context is fully empty for a critical input → ask plainly. This applies to: which entity, which fields (when not "all"), the user's intent itself.

How to ask — prefer AskUserQuestion

When a question must be asked, prefer the AskUserQuestion tool (or its analogue) over writing a numbered list in the response body. Fall back to plain text only if no interactive choice tool is available.

Rules for AskUserQuestion calls in this skill:

• Each call may contain up to 4 questions that are independent of each other (the tool will render them together). Use this to batch related decisions in one round-trip.
• Each question has 2–4 options. The tool auto-adds an "Other" choice for free-form input — never include it manually.
• Mark the recommended option by putting it first with (Recommended) appended to the label.
• Use multiSelect: true for "which fields to include" or "which validators to add" — anything where multiple answers are valid.
• header is a 12-char chip label (e.g. "Variant", "Sub-DTO", "Fields").
• Each option has a description explaining what the choice means or its consequence (one short sentence).

When AskUserQuestion is not the right tool:

• Free-form input where there is no enumerable set of options (e.g. arbitrary class name, arbitrary field rename) — ask in plain text.
• The "single confirmation line" form from principle 2 — that is a plain question with an obvious yes/no, not an enumerated choice.

The screen-driven question lists in Steps 2–5 below are a fallback for case 4. They are NOT a script to execute top-to-bottom. If a question's answer is already determined by principles 1–3, skip the question. Plugin-wizard fidelity is not a goal — generation correctness is. The plugin's UI exists because a screen can show many controls at once; a chat cannot, so the skill must compensate by deciding more on its own.

Step 0 -- Conversation context first (REQUIRED, no tool calls)

Before any MCP call, before any question, re-read the user's prompt and the prior turns of this conversation and extract whatever is already stated. This step costs nothing and prevents the most common failure mode of this skill — asking the user something they already said.

Build a mental checklist of inputs and tick off everything the user has already provided, explicitly or implicitly:

For every input that is explicitly or strongly implicitly answered: mark it as decided and skip the corresponding question in Steps 2–7. Do NOT ask "what entity?" if the user wrote "create DTO for Vet" — Vet is the answer. Do NOT ask "Java record or class?" if the user wrote "make record DtoX" — record is the answer.

For every input that is not answered: defer to the Decision-making principle below — try to derive it from project context first (Step 1), and only then ask.

CRITICAL: ARGUMENTS ≠ user intent. The ARGUMENTS block at the bottom of this prompt is written by the invoking assistant, NOT by the user. It may contain the assistant's own analysis, assumptions, or field-level details that the user never stated. When determining what the user "already said", look ONLY at the actual user messages in the conversation history — never treat ARGUMENTS as a substitute for user input. In particular, do NOT skip asking about sub-DTO shape just because ARGUMENTS describes one.

Step 0 is mental, not a tool call. Do not announce it to the user. Do not write "Step 0 done". Just internalize what the user already said before proceeding to Step 1.

Step 1 -- Gather minimal project context (automatic, no questions)

Call only the MCP tools whose result is actually consumed by a later step. Do not pre-fetch "in case we need it" — every variable here must have a concrete downstream user.

That is the entire Step 1. Do NOT fetch:

• springBootVersion / bootMajor — no branching depends on it
• buildFile — DTO generation does not edit the build file (no dependencies to add); a DTO needs only language constructs
• mainPackage — the DTO's package is derived from the entity's FQN, not from the project root
• list_application_properties_files / propsFile — DTO writes nothing to application.properties
• list_all_domain_entities — needed only if the user did not name an entity in their prompt. Defer to Step 2 as a lazy fallback.
• get_entity_details / list_entity_dtos — depend on knowing the entity, which happens in Step 2. Defer to Step 2.

Derived variables (all from presentDeps and get_project_summary):

• hasLombok = presentDeps contains lombok
• hasJackson = presentDeps contains jackson-databind or jackson-core
• hasValidation = presentDeps contains hibernate-validator or jakarta.validation-api or spring-boot-starter-validation
• canUseRecords = language == JAVA AND jdkVersion >= 16

If multi-module project (multiple modules in get_project_summary): Ask which module to use. Then re-call list_module_dependencies for that module.

Step 2 -- Entity selection

By Step 0 you should already know the entity if the user mentioned it. Most common case: the user wrote "create DTO for Vet" → entity is Vet, skip the question, go straight to the parallel fetch below.

Lazy fallback — only when entity is unknown: call list_all_domain_entities(moduleName) → entities, then ask via AskUserQuestion (options = entity names from the list, max 4; if more than 4 entities, use the 4 most likely candidates based on context and add a note that the user can type a different name via "Other").

This is the only place list_all_domain_entities should be called. If the user named the entity in their prompt, do NOT call it.

After the entity FQN is known, call (in parallel):

list_entity_dtos(entityFqn) for the parent entity is NOT called here — it has no downstream consumer. Sub-entity calls (list_entity_dtos(subEntityFqn)) are still needed for EXIST_CLASS detection in references/sub-dto.md, but they are lazy and per-association, fired only when the user is offered the EXIST_CLASS option. See references/sub-dto.md for the exact place.

DTO name collision is handled by a different lazy call — list_existing_classes(package) — fired right before generation in Step 6 (see anti-hallucination checklist). It is not part of Step 2.

Step 3 -- Attribute selection

Default: include every scalar attribute and every association (with the sub-DTO defaults from references/sub-dto.md). Ask only when context signals that the user wants something narrower.

Decide from context (preferred over asking)

Use the purpose captured in Step 0 to pick a sensible default set:

If the chosen default matches the user's apparent intent, do not ask. Just generate. State the choice in the one-line confirmation form (principle 2) at most.

When to ask

Ask only if:

• The user explicitly said "choose fields" / "ask about fields" / "fine-tune settings", OR
• The entity has many fields and the purpose signal is ambiguous AND the user did not say "use defaults".

When asking, use AskUserQuestion with multiSelect: true:

For each association included, apply the sub-DTO defaults from references/sub-dto.md. Do NOT ask per-association unless the user explicitly requested fine-grained control.

Step 4 -- Variant selection

Apply the Decision-making principle above. The variant is almost always derivable from context — explicit asking should be the exception, not the default.

If language=KOTLIN: route to references/kotlin.md. No question.

If language=JAVA: decide as follows.

Only fall back to the full numbered question when none of the rows above matches AND the user has not said "use defaults". Even then, prefer the "all variants with default marked" format from principle 3 over an iterative question.

Map answer to variant:

• plain → references/java-plain.md
• record → references/java-record.md
• Lombok → references/java-lombok.md

Step 5 -- Variant-specific questions

Follow the variant-specific questions from the selected reference file. Only ask if user did NOT say "all defaults".

Step 6 -- Generate code

1. Determine target path: src/main/{java|kotlin}/{packagePath}/{className}.{java|kt}

2. Follow the Generation Order from the selected reference file.

3. For each step:

   If skeleton (new file):

   • Read the skeleton .md from examples/_skeletons/
   • Apply variable substitutions
   • Use Write tool to create the file

   If fragment:

   • Read the fragment .md from examples/_fragments/
   • Read the Insert Point to know WHERE to insert
   • Use Edit tool to insert code at the specified point
   • Apply variable substitutions

4. Variable substitution rules:

   • {packageName} -> from entity package or user choice
   • {className} -> from user or default {EntityName}Dto
   • Field-level variables -> from entity details
   • NEVER substitute anything not listed in Variables
   • NEVER add imports, methods, or code not in the example
   • FQN handling (CRITICAL): examples contain FQNs (e.g. java.util.Objects, java.util.List, jakarta.validation.constraints.NotNull). When writing the final file, you MUST:
     1. Replace every FQN in the body of the class with its short name (e.g. java.util.Objects.hash(...) -> Objects.hash(...), java.util.List<Integer> -> List<Integer>, @jakarta.validation.constraints.NotNull -> @NotNull).
     2. Collect every FQN you shortened and emit a corresponding import line right after the package statement, sorted, no duplicates.
     3. Types from java.lang (String, Integer, Object, ...) must NOT be imported and must appear as short names.
     4. Classes from the same package as the DTO must NOT be imported.
     5. Javadoc {@link ...} references — UNIFORM: Generate short name + import in {@link …} for every shape: top-level Java class, top-level Java record, nested static class, nested record, and separate-file sub-DTO (NEW_CLASS). There is no asymmetry. Always shorten the entity reference and always add the corresponding import line (unless the entity is in the same package).
     6. Group imports in two blocks separated by ONE blank line:
        • Block 1 — all third-party / project imports together: jakarta.*, com.fasterxml.*, org.springframework.*, org.hibernate.*, project packages, etc. (alphabetical inside the block).
        • (blank line)
        • Block 2 — java.* and javax.* (alphabetical). Do NOT split block 1 into per-package sub-blocks. The final file must contain short names in the body (including every Javadoc {@link …}) and a clean, grouped import block at the top.

5. For sub-DTOs (subDtoType=NEW_CLASS): create a separate file by repeating Steps 6.1-6.4 recursively for the sub-entity.

6. For nested classes (subDtoType=NEW_NESTED_CLASS): add inner class to the parent DTO file, then fill it following the same fragment rules.

7. Nested in a record parent — MANDATORY: when the parent DTO is a Java record, every NEW_NESTED_CLASS association MUST be emitted as a nested public record inside the parent record's body. The skill MUST NOT silently fall back to NEW_CLASS (separate file) just because the record-form fragment looks shorter. The full procedure is in references/java-record.md Step 6 and examples/_fragments/nested-class/java/nested-class.md ("Java record" variant). If those instructions seem ambiguous to you, that is a bug in the skill — fix the docs, do NOT work around it by changing the subDtoType.

Step 7 -- Mapper (automatic when conversion is needed)

Decide whether a mapper is needed based on context, then act:

CRITICAL: Never write manual mapping code (inline toDto/fromDto methods in controllers, services, or anywhere else). If conversion is needed, always delegate to mapper-creator. That skill decides the implementation strategy (MapStruct, Custom mapper, adding dependencies) — this skill just delegates.

Never ask AskUserQuestion for the mapper. If delegating, invoke the mapper-creator skill with the DTO and entity information directly — do not ask the user to confirm the delegation.

Indentation

The skill MUST detect the project's indentation style — never hardcode tabs or spaces. Detection order:

1. .editorconfig at the project root (or any parent of the target file's directory). For Java files, look up the [*.java] or [*] section and read indent_style (tab or space) and indent_size / tab_width.
2. Sample existing Java files in the same package (or the nearest ancestor package that contains Java files). Detect whether the leading whitespace on indented lines uses \t or spaces, and how many.
3. Default to 4-space if neither source is conclusive. (Tabs are also acceptable as a default if the developer explicitly prefers them, but the skill must never silently assume one over the other.)

Whatever style is chosen, apply it uniformly to every line of every generated fragment (fields, constructors, getters/setters, equals/hashCode, toString, nested classes/records). Never mix tabs and spaces inside the same file.

Per-field options

The skill must support the following per-field controls:

• Field rename (fieldNameOverride): the DTO field name can differ from the entity attribute name. Mapper generation still maps it from the original attribute.
• Add validations (extraValidations): the user can add jakarta / hibernate-validator constraints on top of the ones inherited from the entity. The list of allowed constraints depends on the field type — see references/validation.md.
• Remove inherited validations (removedValidations): the user can drop any constraint that came from the entity field.
• Edit annotation parameters (message, min, max, regexp, …): all parameters of every constraint are editable.

These options never appear unless the user explicitly asks for "fine-tune field settings", "per-field validation" or similar. By default the skill just inherits everything from the entity.

Anti-hallucination checklist

Before writing ANY code, verify:

• The code comes from an examples/ file (cite which one)
• Only declared variables were substituted
• No framework API calls were added "from knowledge"
• Import list matches the example exactly
• Method signatures match the example exactly
• No comments or convenience methods were added
• FQNs from examples are shortened in the body AND corresponding import lines were added after package
• Every Javadoc shape — top-level class, top-level record, nested static class, nested record, separate-file sub-DTO — uses short name {@link Pet} AND adds a matching import (unless the entity is in the same package). Uniform rule, no asymmetry.
• If a sub-DTO name (separate file or nested) would collide with an existing class in the target package, auto-suffix with a number (e.g. PetDto1). The skill must do the same — call list_existing_classes(package) before generating, and append 1, 2, … until the name is free.
• When generating a Java record, the inner class for any NEW_NESTED_CLASS association is also a record (not a static class). All component validators are inlined onto the record component parameters (e.g. @NotBlank String firstName), not on separate field declarations.
• When isJavaRecord = true, the skill MUST NOT emit equals(), hashCode(), toString(), mutable setters, or fluent setters — records auto-generate these and these options should not be available for record DTOs.
• Class-level Javadoc is multi-line (/**\n * DTO for {@link …}\n */), never collapsed to one line.
• Getters and setters are multi-line (signature line, indented body, closing brace), never one-liners.
• In mutable=true && fluentSetters=false mode, getters and setters are interleaved (getX, setX, getY, setY, …), not grouped.
• In mutable=true mode, both the no-args constructor and the all-args constructor are emitted.
• The skill does NOT offer "Only ID" as a separate option — this option does not exist (neither for ToOne nor for ToMany). The "association id only" effect is produced by Flat with only the sub-entity id checked, and the skill must implement it that way.
• For collection associations (List<X>, Set<X>), Flat is offered and produces composite plural fields (Set<Integer> specialtyIds, List<Integer> petIds), provided the skill auto-checks the sub-entity scalars.
• Back-reference @ManyToOne fields are filtered out of the attribute list (e.g. Pet.owner is not offered when creating PetDto).
• Imports are grouped with a blank line between jakarta.* / com.* / org.* / java.* blocks.
• Indentation matches the project's detected style (.editorconfig first, then sampling existing files in the same package, then 4-space default). Never hardcode tabs or spaces. Apply uniformly across every fragment in the file.
• @JsonIgnoreProperties is NOT added for Kotlin (Java-only feature)
• Validation annotations use @field: prefix in Kotlin
• Java types are converted to Kotlin types for Kotlin DTOs