dew-method 0.1.0 → 1.0.0

package/package.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/package.json",
   "name": "dew-method",
-  "version": "0.1.0",
+  "version": "1.0.0",
   "description": "Data Engineering Workflow skills packaged as a BMAD-style installable module",
   "keywords": [
     "data-engineering",

package/src/dew-skills/1-analysis/dew-project-brief/SKILL.md

@@ -23,7 +23,10 @@ You may recommend, but you must not decide.
 
 - Do not write a final brief before project scope is clear.
 - Do not treat KPI/source assumptions as facts.
-- If the project has no clear consumer or decision, trigger `HALT-01 — Business Decision Unclear`.
+- Do not ask a long questionnaire when guided interview mode is enabled.
+- Ask one primary question at a time when `guided_interview_mode=true` or `question_batch_size=1`.
+- Every question must include short context, examples/options, and permission to answer `chưa rõ` / `not sure`.
+- If the project has no clear consumer or decision, do guided discovery first; trigger `HALT-01 — Business Decision Unclear` only if the user asks to finalize or proceed past the required gate anyway.
 - Record project-shaping decisions in `.decision-log.md`.
 - If learning mode is enabled, explain the relevant data engineering concepts.
 
@@ -39,7 +42,7 @@ You may recommend, but you must not decide.
 
 3. Load persistent facts from `{workflow.persistent_facts}`.
 
-4. Load `{project-root}/_dew/dew/config.yaml` if present and resolve:
+4. Load `{project-root}/_dew/config.yaml` if present and resolve:
    - `{user_name}`
    - `{project_name}`
    - `{communication_language}`
@@ -47,10 +50,14 @@ You may recommend, but you must not decide.
    - `{planning_artifacts}`
    - `{learning_artifacts}`
    - `{learning_mode}`
+   - `{guided_interview_mode}`
+   - `{question_batch_size}`
+   - `{beginner_prompt_style}`
 
 5. Load:
    - `{workflow.brief_template}`
    - `{workflow.validation_rubric}`
+   - `{workflow.guided_interview_policy}` if present
 
 6. Greet user in `{communication_language}`.
 
@@ -85,11 +92,99 @@ Critique an existing project brief against the validation rubric.
 
 Do not modify the brief unless the user asks.
 
-## Discovery
+## Guided Discovery
 
-Start with a broad project dump.
+When creating a brief, do not start with a broad project dump unless the user explicitly asks for expert mode.
+
+Start with the minimum viable sequence below. Ask exactly one question at a time in guided mode.
+
+### Question 1 — Project idea
+
+Ask:
+
+What data engineering project do you want to build?
+
+Explain that this can be rough. Provide examples:
+
+A. Weather/agriculture advisory data product
+B. Sales KPI dashboard pipeline
+C. Public API to Bronze/Silver/Gold demo
+D. Data quality monitoring project
+E. Not sure yet
+
+After the user answers, summarize the answer in one sentence and continue.
+
+### Question 2 — Target user / consumer
+
+Ask who will use the data product.
+
+Options:
+
+A. Data analyst
+B. Business/internal decision-maker
+C. Farmer/end user/customer
+D. Data engineer/learning portfolio reviewer
+E. Not sure yet
+
+If unclear, help narrow it before continuing.
+
+### Question 3 — Decision or question supported
+
+Ask what decision, action, or question the data product should support.
+
+Options:
+
+A. Monitor performance
+B. Detect risk/anomaly
+C. Compare regions/products/time periods
+D. Recommend an action
+E. Learn/portfolio only for now
+F. Not sure yet
+
+Do not finalize the brief if this remains unclear.
+
+### Question 4 — Data product type
+
+Ask what the expected output is.
+
+Options:
+
+A. Dashboard
+B. Gold data mart
+C. API
+D. Report/notebook
+E. Web app feature
+F. ML/feature table
+G. Not sure yet
+
+Recommend a simple option when the user is a beginner.
+
+### Question 5 — Project type gate
+
+Ask the user to choose:
+
+A. Learning / exploration
+B. Portfolio project
+C. Internal decision-support
+D. Production-grade
+
+Explain the trade-off briefly. Do not continue until the user chooses or asks for guidance.
+
+### Optional Follow-up Questions
+
+After the five required questions, ask follow-ups one at a time only when needed:
+
+- domain
+- known sources
+- known KPIs
+- constraints
+- intended platform/tools
+- learning objectives
+
+## Expert Discovery
+
+If `question_batch_size=all` or the user asks for expert mode, it is acceptable to ask for:
 
-Ask for:
 - project idea
 - target user
 - business or learning goal
@@ -98,16 +193,8 @@ Ask for:
 - known sources
 - known KPIs
 - constraints
-- intended platform/tools, if any
-
-Then classify the project:
-
-A. Learning / exploration  
-B. Portfolio project  
-C. Internal decision-support  
-D. Production-grade  
-
-This is a decision gate. Do not continue until the user chooses.
+- intended platform/tools
+- project type
 
 ## Required Brief Sections
 
@@ -127,6 +214,21 @@ The brief must include:
 - evidence required before architecture
 - next workflow recommendation
 
+## Draft Behavior
+
+If required information is incomplete, create `brief.md` with `status: draft`, not final.
+
+Use frontmatter like:
+
+```yaml
+status: draft
+gate:
+  GATE-00-project-scope: pending
+missing:
+  - target user
+  - business decision
+```
+
 ## Finalize
 
 Before marking the brief as final:

package/src/dew-skills/1-analysis/dew-project-brief/customize.toml

@@ -8,6 +8,7 @@ activation_steps_prepend = []
 activation_steps_append = []
 
 persistent_facts = [
+  "file:{project-root}/_dew/guided-interview-policy.md",
   "file:{project-root}/**/project-context.md",
   "file:{project-root}/**/.decision-log.md",
   "file:{project-root}/**/.learning-log.md",
@@ -17,10 +18,11 @@ on_complete = ""
 
 brief_template = "assets/project-brief-template.md"
 validation_rubric = "assets/project-brief-validation-rubric.md"
+guided_interview_policy = "../../shared/guided-interview-policy.md"
 checklist = "checklist.md"
 
 brief_output_path = "{planning_artifacts}/project-briefs"
 run_folder_pattern = "project-brief-{project_name}-{date}"
 
 decision_log_path = "{planning_artifacts}/.decision-log.md"
-learning_log_path = "{learning_artifacts}/.learning-log.md"
+learning_log_path = "{learning_artifacts}/.learning-log.md"

package/src/dew-skills/module.yaml

@@ -35,6 +35,35 @@ learning_mode:
   default: "true"
   result: "{value}"
 
+guided_interview_mode:
+  prompt:
+    - "Should DEW ask questions one at a time with examples and suggestions?"
+    - "Recommended for beginners and for early planning skills."
+  scope: user
+  default: "true"
+  result: "{value}"
+
+question_batch_size:
+  prompt:
+    - "How many questions should DEW ask at once during discovery?"
+  scope: user
+  default: "1"
+  result: "{value}"
+  single-select:
+    - value: "1"
+      label: "One question at a time - beginner friendly"
+    - value: "3"
+      label: "Small batch - faster but still guided"
+    - value: "all"
+      label: "Full checklist - expert mode"
+
+beginner_prompt_style:
+  prompt:
+    - "Should DEW include examples, choices, and a recommended option when asking questions?"
+  scope: user
+  default: "true"
+  result: "{value}"
+
 planning_artifacts:
   prompt: "Where should planning artifacts be stored?"
   default: "{output_folder}/planning-artifacts"
@@ -115,4 +144,4 @@ agents:
     title: Technical Writer
     icon: "📚"
     team: data-engineering
-    description: "Creates clear documentation, diagrams, learning notes, and reusable patterns."
+    description: "Creates clear documentation, diagrams, learning notes, and reusable patterns."

package/src/dew-skills/shared/guided-interview-policy.md

@@ -0,0 +1,98 @@
+# DEW Guided Interview Policy
+
+## Purpose
+
+Make every DEW skill beginner-friendly by default. Skills must guide users step by step instead of asking for a large form all at once.
+
+## Global Rule
+
+When `guided_interview_mode` is true or `question_batch_size` is `1`, ask exactly one primary question at a time.
+
+Do not ask a 5-10 item questionnaire in a single response unless the user explicitly chooses expert mode.
+
+## Question Format
+
+Every question should include:
+
+1. A short explanation of why the question matters.
+2. 2-5 concrete answer options when possible.
+3. One recommended/default option when reasonable.
+4. Permission to answer `chưa rõ` / `not sure`.
+5. A clear next step after the user answers.
+
+Recommended format:
+
+```markdown
+We need to decide **<concept>** because <reason>.
+
+Question 1/<n>: <single question>
+
+You can answer with:
+A. <option>
+B. <option>
+C. <option>
+D. Not sure yet
+
+Recommended for your case: <option> because <reason>.
+```
+
+## Beginner Mode Behavior
+
+If `user_skill_level` is `beginner`:
+
+- Explain jargon before using it.
+- Prefer examples over abstract definitions.
+- Ask one question at a time.
+- Avoid HALT language unless risk is serious.
+- Use `guided discovery` before declaring a hard halt.
+- Produce draft artifacts with missing fields instead of blocking immediately.
+
+## HALT Behavior
+
+Do not trigger a hard HALT just because the user is unsure.
+
+Use this sequence:
+
+1. Ask one clarifying question.
+2. Offer examples/options.
+3. If still unclear, create a draft artifact with `status: draft` and `missing:` fields.
+4. Trigger HALT only when the user asks to finalize or proceed past a required gate without required evidence.
+
+## Artifact Behavior
+
+When information is incomplete, create a draft artifact instead of a final artifact.
+
+Draft frontmatter example:
+
+```yaml
+status: draft
+gate:
+  GATE-00-project-scope: pending
+missing:
+  - target user
+  - business decision
+```
+
+## Expert Mode
+
+If `question_batch_size` is `all` or the user asks for expert mode:
+
+- It is acceptable to ask a full checklist.
+- Still avoid inventing answers.
+- Clearly label assumptions and missing evidence.
+
+## Applies To
+
+This policy applies to all DEW skills, especially:
+
+- dew-project-brief
+- dew-business-discovery
+- dew-kpi-discovery
+- dew-data-product-prd
+- dew-kpi-feasibility
+- dew-source-discovery
+- dew-source-assessment
+- dew-source-validation
+- dew-requirement-gate
+- dew-data-architecture
+- dew-create-data-story