Chatbot Builder — Chatbot Construction Pipeline
An agent team collaborates to build a chatbot through persona design > conversation design > NLU > integration > testing.
Execution Mode
Agent Team — Five agents communicate directly via SendMessage and perform cross-validation.
Agent Composition
| Agent |
File |
Role |
Type |
| persona-architect |
.claude/agents/persona-architect.md |
Bot persona design |
general-purpose |
| conversation-designer |
.claude/agents/conversation-designer.md |
Conversation scenario design |
general-purpose |
| nlu-developer |
.claude/agents/nlu-developer.md |
NLU pipeline implementation |
general-purpose |
| integration-engineer |
.claude/agents/integration-engineer.md |
Channel integration, deployment |
general-purpose |
| dialog-tester |
.claude/agents/dialog-tester.md |
Quality verification, testing |
general-purpose |
Workflow
Phase 1: Preparation (performed directly by the orchestrator)
- Extract the following from user input:
- Chatbot purpose: Customer service/FAQ/reservations/orders/consultation, etc.
- Target users: Age, digital literacy, expectation level
- Integration channels: Slack/KakaoTalk/Telegram/Web/multi-channel
- Domain knowledge: Business rules, FAQ list, product information, etc.
- Constraints (optional): Technology stack, response time requirements
- Create the
_workspace/ directory at the project root
- Organize the input and save it to
_workspace/00_input.md
- Create the
_workspace/src/ directory
- If pre-existing files are available, copy them to
_workspace/ and skip the corresponding phase
- Determine the execution mode based on the scope of the request (see "Execution Modes by Request Scope" below)
Phase 2: Team Assembly and Execution
| Order |
Task |
Owner |
Dependencies |
Deliverable |
| 1 |
Persona design |
persona |
None |
_workspace/01_persona_spec.md |
| 2 |
Conversation design |
designer |
Task 1 |
_workspace/02_conversation_design.md |
| 3a |
NLU implementation |
nlu-dev |
Task 2 |
_workspace/03_nlu_config.md + src/ |
| 3b |
Integration design |
integrator |
Task 2 |
_workspace/04_integration_spec.md + src/ |
| 4 |
Testing |
tester |
Tasks 3a, 3b |
_workspace/05_test_report.md |
Tasks 3a (NLU) and 3b (integration) run in parallel.
Inter-agent communication flow:
- persona completes > passes tone and manner guide to designer, passes domain keywords to nlu-dev
- designer completes > passes intent/entity catalog to nlu-dev, passes external integration flows to integrator
- nlu-dev completes > passes NLU interface to integrator, passes test data to tester
- integrator completes > passes test environment info to tester
- tester cross-validates all deliverables. On CRITICAL findings, requests corrections from the relevant agent > rework > re-verification (up to 2 rounds)
Phase 3: Integration and Final Deliverables
Finalize deliverables based on the tester's report:
- Verify all files in
_workspace/ and src/ code
- Confirm that all CRITICAL findings have been resolved
- Report the final summary to the user
Execution Modes by Request Scope
| User Request Pattern |
Execution Mode |
Agents Deployed |
| "Build me a chatbot", "full build" |
Full pipeline |
All 5 agents |
| "Just do the conversation design", "write scenarios" |
Design mode |
persona + designer |
| "Just develop the NLU" (design complete) |
NLU mode |
nlu-dev + tester |
| "Integrate chatbot with KakaoTalk" (implementation complete) |
Integration mode |
integrator + tester |
| "Test the chatbot" (implementation complete) |
Test mode |
tester only |
Reusing existing files: If the user provides an existing conversation design document or NLU configuration, copy those files to _workspace/ and skip the corresponding steps.
Data Transfer Protocol
| Strategy |
Method |
Purpose |
| File-based |
_workspace/ directory |
Design documents and configuration sharing |
| Message-based |
SendMessage |
Real-time key information transfer, correction requests |
| Code-based |
_workspace/src/ |
Executable source code |
Error Handling
| Error Type |
Strategy |
| Insufficient domain knowledge |
Request additional FAQ list/business information from user, supplement with web search |
| Channel API changes |
Verify latest API documentation via WebFetch, then update integration code |
| NLU accuracy below threshold |
Augment training data > redesign prompts > strengthen fallback, in that order |
| Agent failure |
Retry once > if still failing, proceed without that deliverable and note in report |
Test Scenarios
Normal Flow
Prompt: "Build a KakaoTalk chatbot for cafe order processing. It should support menu browsing, ordering, and payment guidance"
Expected result:
- Persona: Bright, friendly cafe staff character, formal tone, active emoji usage
- Conversation design: Menu browsing/ordering/payment guidance/business hours intents, slots (menu name/quantity/options)
- NLU: LLM prompt-based intent classification, menu name entity dictionary construction
- Integration: KakaoTalk chatbot API integration, card-type messages for menu display
- Testing: Order flow happy path, fallback for unavailable menu items, multi-turn ordering
Existing File Reuse Flow
Prompt: "I have this conversation design document; implement the NLU and integrate it with KakaoTalk" + design document attached
Expected result:
- Copy existing design document to
_workspace/02_conversation_design.md
- Skip persona and designer; deploy nlu-dev + integrator + tester
- Implement NLU based on existing intent/entity catalog
Error Flow
Prompt: "Build me a chatbot" (purpose, channel unclear)
Expected result:
- persona proposes a generic persona and requests purpose confirmation from user
- Proceed with remaining pipeline after purpose is confirmed
- Note in report: "Channel undetermined — web widget applied as default"
Agent Extension Skills
Extension skills that enhance agent domain expertise:
| Skill |
File |
Target Agent |
Role |
| intent-taxonomy-builder |
.claude/skills/intent-taxonomy-builder/skill.md |
nlu-developer, conversation-designer |
Intent taxonomy design, entity-slot mapping, training data generation, confusion matrix |
| conversation-flow-validator |
.claude/skills/conversation-flow-validator/skill.md |
dialog-tester, conversation-designer |
Dialog flow defect detection, Happy/Sad/Edge testing, fallback hierarchy, quality metrics |
1---2name: chatbot-builder3description: Full pipeline where an agent team collaborates to build a chatbot system. Use this skill for requests like 'build me a chatbot', 'conversational bot development', 'customer service bot', 'FAQ chatbot', 'KakaoTalk chatbot', 'Slack bot', 'auto-response system', 'conversational AI', 'chatbot design', and other chatbot construction tasks. Also supports design-only mode when only conversation design is needed. Note: voice assistants (Alexa/Google Home), real-time voice call bots, and video chatbots are outside the scope of this skill.4---56# Chatbot Builder — Chatbot Construction Pipeline78An agent team collaborates to build a chatbot through persona design > conversation design > NLU > integration > testing.910## Execution Mode1112**Agent Team** — Five agents communicate directly via SendMessage and perform cross-validation.1314## Agent Composition1516| Agent | File | Role | Type |17|-------|------|------|------|18| persona-architect | `.claude/agents/persona-architect.md` | Bot persona design | general-purpose |19| conversation-designer | `.claude/agents/conversation-designer.md` | Conversation scenario design | general-purpose |20| nlu-developer | `.claude/agents/nlu-developer.md` | NLU pipeline implementation | general-purpose |21| integration-engineer | `.claude/agents/integration-engineer.md` | Channel integration, deployment | general-purpose |22| dialog-tester | `.claude/agents/dialog-tester.md` | Quality verification, testing | general-purpose |2324## Workflow2526### Phase 1: Preparation (performed directly by the orchestrator)27281. Extract the following from user input:29 - **Chatbot purpose**: Customer service/FAQ/reservations/orders/consultation, etc.30 - **Target users**: Age, digital literacy, expectation level31 - **Integration channels**: Slack/KakaoTalk/Telegram/Web/multi-channel32 - **Domain knowledge**: Business rules, FAQ list, product information, etc.33 - **Constraints** (optional): Technology stack, response time requirements342. Create the `_workspace/` directory at the project root353. Organize the input and save it to `_workspace/00_input.md`364. Create the `_workspace/src/` directory375. If pre-existing files are available, copy them to `_workspace/` and skip the corresponding phase386. **Determine the execution mode** based on the scope of the request (see "Execution Modes by Request Scope" below)3940### Phase 2: Team Assembly and Execution4142| Order | Task | Owner | Dependencies | Deliverable |43|-------|------|-------|-------------|-------------|44| 1 | Persona design | persona | None | `_workspace/01_persona_spec.md` |45| 2 | Conversation design | designer | Task 1 | `_workspace/02_conversation_design.md` |46| 3a | NLU implementation | nlu-dev | Task 2 | `_workspace/03_nlu_config.md` + `src/` |47| 3b | Integration design | integrator | Task 2 | `_workspace/04_integration_spec.md` + `src/` |48| 4 | Testing | tester | Tasks 3a, 3b | `_workspace/05_test_report.md` |4950Tasks 3a (NLU) and 3b (integration) run **in parallel**.5152**Inter-agent communication flow:**53- persona completes > passes tone and manner guide to designer, passes domain keywords to nlu-dev54- designer completes > passes intent/entity catalog to nlu-dev, passes external integration flows to integrator55- nlu-dev completes > passes NLU interface to integrator, passes test data to tester56- integrator completes > passes test environment info to tester57- tester cross-validates all deliverables. On CRITICAL findings, requests corrections from the relevant agent > rework > re-verification (up to 2 rounds)5859### Phase 3: Integration and Final Deliverables6061Finalize deliverables based on the tester's report:62631. Verify all files in `_workspace/` and `src/` code642. Confirm that all CRITICAL findings have been resolved653. Report the final summary to the user6667## Execution Modes by Request Scope6869| User Request Pattern | Execution Mode | Agents Deployed |70|---------------------|---------------|----------------|71| "Build me a chatbot", "full build" | **Full pipeline** | All 5 agents |72| "Just do the conversation design", "write scenarios" | **Design mode** | persona + designer |73| "Just develop the NLU" (design complete) | **NLU mode** | nlu-dev + tester |74| "Integrate chatbot with KakaoTalk" (implementation complete) | **Integration mode** | integrator + tester |75| "Test the chatbot" (implementation complete) | **Test mode** | tester only |7677**Reusing existing files**: If the user provides an existing conversation design document or NLU configuration, copy those files to `_workspace/` and skip the corresponding steps.7879## Data Transfer Protocol8081| Strategy | Method | Purpose |82|----------|--------|---------|83| File-based | `_workspace/` directory | Design documents and configuration sharing |84| Message-based | SendMessage | Real-time key information transfer, correction requests |85| Code-based | `_workspace/src/` | Executable source code |8687## Error Handling8889| Error Type | Strategy |90|-----------|----------|91| Insufficient domain knowledge | Request additional FAQ list/business information from user, supplement with web search |92| Channel API changes | Verify latest API documentation via WebFetch, then update integration code |93| NLU accuracy below threshold | Augment training data > redesign prompts > strengthen fallback, in that order |94| Agent failure | Retry once > if still failing, proceed without that deliverable and note in report |9596## Test Scenarios9798### Normal Flow99**Prompt**: "Build a KakaoTalk chatbot for cafe order processing. It should support menu browsing, ordering, and payment guidance"100**Expected result**:101- Persona: Bright, friendly cafe staff character, formal tone, active emoji usage102- Conversation design: Menu browsing/ordering/payment guidance/business hours intents, slots (menu name/quantity/options)103- NLU: LLM prompt-based intent classification, menu name entity dictionary construction104- Integration: KakaoTalk chatbot API integration, card-type messages for menu display105- Testing: Order flow happy path, fallback for unavailable menu items, multi-turn ordering106107### Existing File Reuse Flow108**Prompt**: "I have this conversation design document; implement the NLU and integrate it with KakaoTalk" + design document attached109**Expected result**:110- Copy existing design document to `_workspace/02_conversation_design.md`111- Skip persona and designer; deploy nlu-dev + integrator + tester112- Implement NLU based on existing intent/entity catalog113114### Error Flow115**Prompt**: "Build me a chatbot" (purpose, channel unclear)116**Expected result**:117- persona proposes a generic persona and requests purpose confirmation from user118- Proceed with remaining pipeline after purpose is confirmed119- Note in report: "Channel undetermined — web widget applied as default"120121## Agent Extension Skills122123Extension skills that enhance agent domain expertise:124125| Skill | File | Target Agent | Role |126|-------|------|-------------|------|127| intent-taxonomy-builder | `.claude/skills/intent-taxonomy-builder/skill.md` | nlu-developer, conversation-designer | Intent taxonomy design, entity-slot mapping, training data generation, confusion matrix |128| conversation-flow-validator | `.claude/skills/conversation-flow-validator/skill.md` | dialog-tester, conversation-designer | Dialog flow defect detection, Happy/Sad/Edge testing, fallback hierarchy, quality metrics |
revfactory