diff --git a/.claude/commands/mgw/ask.md b/.claude/commands/mgw/ask.md
new file mode 100644
index 0000000..c8b06f3
--- /dev/null
+++ b/.claude/commands/mgw/ask.md
@@ -0,0 +1,368 @@
+---
+name: mgw:ask
+description: Route a question/observation — classify as in-scope, adjacent, separate, duplicate, or out-of-scope
+argument-hint: "<question>"
+allowed-tools:
+  - Bash
+  - Read
+  - Task
+  - AskUserQuestion
+---
+
+<objective>
+During milestone execution, observations arise that need classification: is this
+in-scope for the current issue, adjacent to a different issue, a separate concern,
+a duplicate, or out-of-scope entirely?
+
+/mgw:ask spawns a general-purpose agent with full project context (milestone, all
+issues, active state, recent git diff) that classifies the question and recommends
+an action. Read-only for the orchestrator — the agent reads state and code, MGW
+presents the structured result.
+</objective>
+
+<execution_context>
+@~/.claude/commands/mgw/workflows/state.md
+@~/.claude/commands/mgw/workflows/github.md
+@~/.claude/commands/mgw/workflows/gsd.md
+@~/.claude/commands/mgw/workflows/validation.md
+</execution_context>
+
+<context>
+Question text: $ARGUMENTS
+
+Active issues context: .mgw/active/ (current work state)
+Project context: .mgw/project.json (milestone + all issues)
+</context>
+
+<process>
+
+<step name="validate_input">
+**Validate question text provided:**
+
+Parse $ARGUMENTS for the question text. If missing or empty:
+```
+AskUserQuestion(
+  header: "Question Required",
+  question: "What question or observation do you want to classify?",
+  followUp: "Enter a question or observation (e.g., 'The slug generation doesn't handle unicode characters')"
+)
+```
+Store as $QUESTION.
+</step>
+
+<step name="load_project_context">
+**Load project.json for milestone and issue context:**
+
+```bash
+REPO_ROOT=$(git rev-parse --show-toplevel)
+MGW_DIR="${REPO_ROOT}/.mgw"
+
+# Load project state
+PROJECT_JSON=""
+MILESTONE_CONTEXT=""
+ALL_ISSUES_CONTEXT=""
+
+if [ -f "${MGW_DIR}/project.json" ]; then
+  PROJECT_JSON=$(cat "${MGW_DIR}/project.json")
+
+  # Extract current milestone info
+  MILESTONE_CONTEXT=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx < len(p['milestones']):
+    m = p['milestones'][idx]
+    print(f\"Milestone: {m['name']}\")
+    if m.get('description'):
+        print(f\"Description: {m['description']}\")
+else:
+    print('No active milestone')
+" 2>/dev/null || echo "No project initialized")
+
+  # Extract all issues in current milestone with titles, bodies, scopes, labels
+  ALL_ISSUES_CONTEXT=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx >= len(p['milestones']):
+    print('No issues found')
+    sys.exit(0)
+m = p['milestones'][idx]
+for issue in m.get('issues', []):
+    num = issue.get('github_number', '?')
+    title = issue.get('title', 'untitled')
+    stage = issue.get('pipeline_stage', 'new')
+    labels = ', '.join(issue.get('labels', []))
+    phase = issue.get('phase_name', '')
+    scope = issue.get('scope', '')
+    print(f'#{num} [{stage}] {title}')
+    if phase:
+        print(f'  Phase: {phase}')
+    if labels:
+        print(f'  Labels: {labels}')
+    if scope:
+        print(f'  Scope: {scope}')
+    print()
+" 2>/dev/null || echo "")
+else
+  MILESTONE_CONTEXT="No project initialized"
+  ALL_ISSUES_CONTEXT=""
+fi
+```
+</step>
+
+<step name="load_active_state">
+**Load active issue state from .mgw/active/:**
+
+```bash
+ACTIVE_STATE=""
+ACTIVE_ISSUE_NUMBER=""
+
+ACTIVE_FILES=$(ls "${MGW_DIR}/active/"*.json 2>/dev/null)
+if [ -n "$ACTIVE_FILES" ]; then
+  for f in ${ACTIVE_FILES}; do
+    ISSUE_DATA=$(cat "$f")
+    ISSUE_NUM=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('issue',{}).get('number','?'))")
+    ISSUE_TITLE=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('issue',{}).get('title','untitled'))")
+    PIPELINE_STAGE=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('pipeline_stage','unknown'))")
+    ACTIVE_STATE="${ACTIVE_STATE}#${ISSUE_NUM} [${PIPELINE_STAGE}] ${ISSUE_TITLE}\n"
+    # Track the most recently active issue (last in list)
+    ACTIVE_ISSUE_NUMBER="$ISSUE_NUM"
+  done
+else
+  ACTIVE_STATE="No active issues"
+fi
+```
+</step>
+
+<step name="gather_git_diff">
+**Gather recent git diff for change context:**
+
+```bash
+# Get recent changes (staged + unstaged, limited to keep prompt reasonable)
+RECENT_DIFF=$(git diff HEAD --stat 2>/dev/null | head -30 || echo "No changes")
+CURRENT_BRANCH=$(git branch --show-current 2>/dev/null || echo "unknown")
+```
+</step>
+
+<step name="fetch_issue_details">
+**Fetch full issue bodies from GitHub for matching context:**
+
+If project.json has issues, fetch their bodies for the agent to compare against:
+
+```bash
+ISSUE_BODIES=""
+if [ -n "$PROJECT_JSON" ]; then
+  ISSUE_NUMBERS=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx < len(p['milestones']):
+    m = p['milestones'][idx]
+    for issue in m.get('issues', []):
+        print(issue.get('github_number', ''))
+" 2>/dev/null)
+
+  for NUM in $ISSUE_NUMBERS; do
+    if [ -n "$NUM" ]; then
+      BODY=$(gh issue view "$NUM" --json number,title,body -q '"\(.number)|\(.title)|\(.body)"' 2>/dev/null || echo "")
+      if [ -n "$BODY" ]; then
+        ISSUE_BODIES="${ISSUE_BODIES}${BODY}\n---\n"
+      fi
+    fi
+  done
+fi
+```
+</step>
+
+<step name="spawn_classification_agent">
+**Spawn general-purpose agent for question classification:**
+
+```
+Task(
+  prompt="
+<files_to_read>
+- ./CLAUDE.md (Project instructions — if exists, follow all guidelines)
+- .agents/skills/ (Project skills — if dir exists, list skills, read SKILL.md for each, follow relevant rules)
+</files_to_read>
+
+You are a question classification agent for the MGW pipeline. Your job is to
+classify a question or observation against the current project context and
+recommend an action.
+
+<question>
+${QUESTION}
+</question>
+
+<current_milestone>
+${MILESTONE_CONTEXT}
+</current_milestone>
+
+<milestone_issues>
+${ALL_ISSUES_CONTEXT}
+</milestone_issues>
+
+<issue_bodies>
+${ISSUE_BODIES}
+</issue_bodies>
+
+<active_work>
+Current branch: ${CURRENT_BRANCH}
+Active issues in .mgw/:
+${ACTIVE_STATE}
+</active_work>
+
+<recent_changes>
+${RECENT_DIFF}
+</recent_changes>
+
+<classification_rules>
+
+Classify the question into exactly ONE of these categories:
+
+| Category | Criteria | Action |
+|----------|----------|--------|
+| In-scope | Directly relates to the current active issue being worked on | Include in current work — no new issue needed |
+| Adjacent | Relates to a DIFFERENT issue in the same milestone | Note it on that issue — suggest posting a comment |
+| Separate | Doesn't match any open issue in the milestone | Suggest filing a new issue with a title |
+| Duplicate | Matches an existing issue (same root cause or fix) | Point to the existing issue |
+| Out-of-scope | Beyond the current milestone entirely | Note for future planning |
+
+Decision process:
+1. Read the question carefully
+2. Compare against each issue title, body, and scope in the milestone
+3. Check if it relates to current active work (branch, diff, active state)
+4. Look for keyword/concept overlap with existing issues
+5. If no match, determine if it fits the milestone's theme or is out-of-scope
+
+</classification_rules>
+
+<output_format>
+Return a structured classification report:
+
+## Classification: ${CATEGORY}
+
+### Analysis
+- What the question is about (1-2 sentences)
+- Why this classification was chosen (1-2 sentences)
+
+### Related Issue
+- Issue number and title (if adjacent or duplicate)
+- Or 'N/A — current work' (if in-scope)
+- Or 'No matching issue' (if separate or out-of-scope)
+
+### Recommendation
+- Specific actionable next step
+- If 'separate': suggest an issue title and brief body
+- If 'adjacent': suggest a comment to post on the related issue
+- If 'in-scope': note what to include in current work
+- If 'duplicate': point to the matching issue
+- If 'out-of-scope': note for future milestone planning
+</output_format>
+",
+  subagent_type="general-purpose",
+  description="Classify question: ${QUESTION}"
+)
+```
+</step>
+
+<step name="present_result">
+**Present the classification result to the user:**
+
+Display the agent's report, then present the MGW action banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ MGW ► QUESTION ROUTING
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+${classification_report_from_agent}
+
+───────────────────────────────────────────────────────
+```
+</step>
+
+<step name="offer_actions">
+**Offer follow-up actions based on classification:**
+
+**If "Separate":**
+```
+AskUserQuestion(
+  header: "File New Issue?",
+  question: "Create a new issue from this observation?",
+  options: [
+    { label: "Yes", description: "File the suggested issue via /mgw:issue" },
+    { label: "No", description: "Note it and continue current work" }
+  ]
+)
+```
+
+If user says "Yes":
+```
+Suggested issue ready. To file it:
+
+  gh issue create --title "${suggested_title}" --body "${suggested_body}"
+
+Or use /mgw:project to add it to a future milestone.
+```
+
+**If "Adjacent":**
+```
+AskUserQuestion(
+  header: "Post Comment?",
+  question: "Post a note on the related issue #${related_number}?",
+  options: [
+    { label: "Yes", description: "Post observation as a comment on #${related_number}" },
+    { label: "No", description: "Note it and continue current work" }
+  ]
+)
+```
+
+If user says "Yes":
+```bash
+gh issue comment ${related_number} --body "> **MGW** · \`observation\` · $(node ~/.claude/get-shit-done/bin/gsd-tools.cjs current-timestamp --raw)
+
+Observation noted during work on #${ACTIVE_ISSUE_NUMBER}:
+
+${QUESTION}"
+```
+
+Report: "Comment posted on #${related_number}."
+
+**If "In-scope":**
+```
+This relates to your current work on #${ACTIVE_ISSUE_NUMBER}.
+No action needed — include it in your current implementation.
+```
+
+**If "Duplicate":**
+```
+This appears to duplicate #${duplicate_number} — ${duplicate_title}.
+No new issue needed. Check that issue for existing progress.
+```
+
+**If "Out-of-scope":**
+```
+This is beyond the current milestone scope.
+Consider adding it to a future milestone or filing it for backlog:
+
+  gh issue create --title "${suggested_title}" --label "backlog"
+```
+</step>
+
+</process>
+
+<success_criteria>
+- [ ] Question text parsed from $ARGUMENTS (or prompted)
+- [ ] project.json loaded for milestone + issue context
+- [ ] .mgw/active/ state loaded for current work context
+- [ ] Recent git diff gathered for change context
+- [ ] Issue bodies fetched from GitHub for comparison
+- [ ] Classification agent spawned with full context
+- [ ] Classification returned: in-scope, adjacent, separate, duplicate, or out-of-scope
+- [ ] Related issue identified (if adjacent or duplicate)
+- [ ] Actionable recommendation provided
+- [ ] Follow-up action offered (file issue, post comment, etc.)
+- [ ] Delegation boundary respected: agent reads code/state, MGW presents results
+</success_criteria>
+</output>
diff --git a/.claude/commands/mgw/help.md b/.claude/commands/mgw/help.md
index b15dbba..6d8f6e1 100644
--- a/.claude/commands/mgw/help.md
+++ b/.claude/commands/mgw/help.md
@@ -53,6 +53,10 @@ COMMANDS
   /mgw:pr [number] [--base b]  Create PR from GSD artifacts + issue context
   /mgw:link <ref> <ref>        Cross-reference issues/PRs/branches
 
+  Query
+  ─────
+  /mgw:ask <question>          Route a question — in-scope, adjacent, separate, duplicate, out-of-scope
+
   Maintenance
   ───────────
   /mgw:sync                    Reconcile .mgw/ state with GitHub
diff --git a/.claude/commands/mgw/workflows/github.md b/.claude/commands/mgw/workflows/github.md
index 9dd46c7..f28ecc6 100644
--- a/.claude/commands/mgw/workflows/github.md
+++ b/.claude/commands/mgw/workflows/github.md
@@ -259,9 +259,9 @@ Release is created as draft — user reviews and publishes manually.
 
 | Section | Referenced By |
 |---------|-------------|
-| Issue Operations | issue.md, run.md, issues.md, sync.md, milestone.md, next.md |
+| Issue Operations | issue.md, run.md, issues.md, sync.md, milestone.md, next.md, ask.md |
 | Comment Operations | issue.md (triage snapshot), run.md (pre-flight check), sync.md (drift), review.md |
-| Issue Mutations | issue.md, update.md, run.md, init.md, milestone.md |
+| Issue Mutations | issue.md, update.md, run.md, init.md, milestone.md, ask.md |
 | Milestone Operations | project.md, milestone.md |
 | Dependency Labels | project.md |
 | Phase Labels | project.md |
diff --git a/.claude/commands/mgw/workflows/gsd.md b/.claude/commands/mgw/workflows/gsd.md
index 30b243d..fed852c 100644
--- a/.claude/commands/mgw/workflows/gsd.md
+++ b/.claude/commands/mgw/workflows/gsd.md
@@ -218,6 +218,42 @@ ARTIFACT_CHECK=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs verify artifacts
 KEYLINK_CHECK=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs verify key-links "${PLAN_PATH}" 2>/dev/null || echo '{"passed":true}')
 ```
 
+## Question Classification Agent Pattern
+
+Used by `/mgw:ask` to classify questions/observations during milestone execution.
+Spawns a general-purpose agent with full project context — milestone, all issues,
+active state, and recent git diff — to determine if a question is in-scope,
+adjacent, separate, duplicate, or out-of-scope.
+
+```
+Task(
+  prompt="
+    <files_to_read>
+    - ./CLAUDE.md (Project instructions — if exists, follow all guidelines)
+    - .agents/skills/ (Project skills — if dir exists, list skills, read SKILL.md for each, follow relevant rules)
+    </files_to_read>
+
+    You are a question classification agent for the MGW pipeline.
+
+    <question>${QUESTION}</question>
+    <current_milestone>${MILESTONE_CONTEXT}</current_milestone>
+    <milestone_issues>${ALL_ISSUES_CONTEXT}</milestone_issues>
+    <issue_bodies>${ISSUE_BODIES}</issue_bodies>
+    <active_work>${ACTIVE_STATE}</active_work>
+    <recent_changes>${RECENT_DIFF}</recent_changes>
+
+    Classify into: in-scope | adjacent | separate | duplicate | out-of-scope
+    Return: classification, analysis, related issue, recommendation.
+  ",
+  subagent_type="general-purpose",
+  description="Classify question: ${QUESTION}"
+)
+```
+
+The agent is read-only (general-purpose, no code execution). It reads project state
+and codebase to classify, then MGW presents the result and offers follow-up actions
+(file new issue, post comment on related issue, etc.).
+
 ## Anti-Patterns
 
 - **NEVER** use Skill invocation from within a Task() agent — Skills don't resolve inside subagents
@@ -231,9 +267,10 @@ KEYLINK_CHECK=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs verify key-links
 
 | Pattern | Referenced By |
 |---------|-------------|
-| Standard spawn template | run.md, issue.md, pr.md, review.md |
+| Standard spawn template | run.md, issue.md, pr.md, ask.md, review.md |
 | Comment classification | run.md (pre-flight), review.md (standalone) |
 | Quick pipeline | run.md |
 | Milestone pipeline | run.md |
+| Question classification | ask.md |
 | Model resolution | run.md |
-| Utility patterns | run.md, pr.md, issue.md, sync.md, link.md, update.md |
+| Utility patterns | run.md, pr.md, issue.md, sync.md, link.md, update.md, ask.md |
diff --git a/.claude/commands/mgw/workflows/state.md b/.claude/commands/mgw/workflows/state.md
index 016e4dc..34ff37f 100644
--- a/.claude/commands/mgw/workflows/state.md
+++ b/.claude/commands/mgw/workflows/state.md
@@ -319,11 +319,11 @@ Only advance if ALL issues in current milestone completed successfully.
 
 | Pattern | Referenced By |
 |---------|-------------|
-| validate_and_load | init.md, issue.md, run.md, update.md, link.md, pr.md, sync.md, milestone.md |
+| validate_and_load | init.md, issue.md, run.md, update.md, link.md, pr.md, sync.md, milestone.md, ask.md |
 | Per-issue staleness | run.md, issue.md, update.md |
 | Batch staleness | sync.md (full reconciliation), milestone.md |
 | Comment tracking | issue.md (populate), run.md (pre-flight check), sync.md (drift detection) |
 | Issue state schema | issue.md, run.md, update.md, sync.md |
 | Cross-refs schema | link.md, run.md, pr.md, sync.md |
 | Slug generation | issue.md, run.md |
-| Project state | milestone.md, next.md |
+| Project state | milestone.md, next.md, ask.md |
diff --git a/.claude/commands/mgw/workflows/validation.md b/.claude/commands/mgw/workflows/validation.md
index 4a182df..fea1adb 100644
--- a/.claude/commands/mgw/workflows/validation.md
+++ b/.claude/commands/mgw/workflows/validation.md
@@ -135,6 +135,7 @@ This rule applies to ALL MGW commands:
 | issue.md | Spawns analysis agent — never analyzes code itself |
 | review.md | Spawns comment classification agent — reads comments, not code |
 | pr.md | Spawns PR body builder — never reads code for description |
+| ask.md | Spawns classification agent — never analyzes code itself |
 | sync.md | Reads state + GitHub API only — never reads code |
 | update.md | Reads state only — posts comments |
 | link.md | Reads/writes cross-refs only |
diff --git a/commands/ask.md b/commands/ask.md
new file mode 100644
index 0000000..c8b06f3
--- /dev/null
+++ b/commands/ask.md
@@ -0,0 +1,368 @@
+---
+name: mgw:ask
+description: Route a question/observation — classify as in-scope, adjacent, separate, duplicate, or out-of-scope
+argument-hint: "<question>"
+allowed-tools:
+  - Bash
+  - Read
+  - Task
+  - AskUserQuestion
+---
+
+<objective>
+During milestone execution, observations arise that need classification: is this
+in-scope for the current issue, adjacent to a different issue, a separate concern,
+a duplicate, or out-of-scope entirely?
+
+/mgw:ask spawns a general-purpose agent with full project context (milestone, all
+issues, active state, recent git diff) that classifies the question and recommends
+an action. Read-only for the orchestrator — the agent reads state and code, MGW
+presents the structured result.
+</objective>
+
+<execution_context>
+@~/.claude/commands/mgw/workflows/state.md
+@~/.claude/commands/mgw/workflows/github.md
+@~/.claude/commands/mgw/workflows/gsd.md
+@~/.claude/commands/mgw/workflows/validation.md
+</execution_context>
+
+<context>
+Question text: $ARGUMENTS
+
+Active issues context: .mgw/active/ (current work state)
+Project context: .mgw/project.json (milestone + all issues)
+</context>
+
+<process>
+
+<step name="validate_input">
+**Validate question text provided:**
+
+Parse $ARGUMENTS for the question text. If missing or empty:
+```
+AskUserQuestion(
+  header: "Question Required",
+  question: "What question or observation do you want to classify?",
+  followUp: "Enter a question or observation (e.g., 'The slug generation doesn't handle unicode characters')"
+)
+```
+Store as $QUESTION.
+</step>
+
+<step name="load_project_context">
+**Load project.json for milestone and issue context:**
+
+```bash
+REPO_ROOT=$(git rev-parse --show-toplevel)
+MGW_DIR="${REPO_ROOT}/.mgw"
+
+# Load project state
+PROJECT_JSON=""
+MILESTONE_CONTEXT=""
+ALL_ISSUES_CONTEXT=""
+
+if [ -f "${MGW_DIR}/project.json" ]; then
+  PROJECT_JSON=$(cat "${MGW_DIR}/project.json")
+
+  # Extract current milestone info
+  MILESTONE_CONTEXT=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx < len(p['milestones']):
+    m = p['milestones'][idx]
+    print(f\"Milestone: {m['name']}\")
+    if m.get('description'):
+        print(f\"Description: {m['description']}\")
+else:
+    print('No active milestone')
+" 2>/dev/null || echo "No project initialized")
+
+  # Extract all issues in current milestone with titles, bodies, scopes, labels
+  ALL_ISSUES_CONTEXT=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx >= len(p['milestones']):
+    print('No issues found')
+    sys.exit(0)
+m = p['milestones'][idx]
+for issue in m.get('issues', []):
+    num = issue.get('github_number', '?')
+    title = issue.get('title', 'untitled')
+    stage = issue.get('pipeline_stage', 'new')
+    labels = ', '.join(issue.get('labels', []))
+    phase = issue.get('phase_name', '')
+    scope = issue.get('scope', '')
+    print(f'#{num} [{stage}] {title}')
+    if phase:
+        print(f'  Phase: {phase}')
+    if labels:
+        print(f'  Labels: {labels}')
+    if scope:
+        print(f'  Scope: {scope}')
+    print()
+" 2>/dev/null || echo "")
+else
+  MILESTONE_CONTEXT="No project initialized"
+  ALL_ISSUES_CONTEXT=""
+fi
+```
+</step>
+
+<step name="load_active_state">
+**Load active issue state from .mgw/active/:**
+
+```bash
+ACTIVE_STATE=""
+ACTIVE_ISSUE_NUMBER=""
+
+ACTIVE_FILES=$(ls "${MGW_DIR}/active/"*.json 2>/dev/null)
+if [ -n "$ACTIVE_FILES" ]; then
+  for f in ${ACTIVE_FILES}; do
+    ISSUE_DATA=$(cat "$f")
+    ISSUE_NUM=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('issue',{}).get('number','?'))")
+    ISSUE_TITLE=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('issue',{}).get('title','untitled'))")
+    PIPELINE_STAGE=$(echo "$ISSUE_DATA" | python3 -c "import json,sys; print(json.load(sys.stdin).get('pipeline_stage','unknown'))")
+    ACTIVE_STATE="${ACTIVE_STATE}#${ISSUE_NUM} [${PIPELINE_STAGE}] ${ISSUE_TITLE}\n"
+    # Track the most recently active issue (last in list)
+    ACTIVE_ISSUE_NUMBER="$ISSUE_NUM"
+  done
+else
+  ACTIVE_STATE="No active issues"
+fi
+```
+</step>
+
+<step name="gather_git_diff">
+**Gather recent git diff for change context:**
+
+```bash
+# Get recent changes (staged + unstaged, limited to keep prompt reasonable)
+RECENT_DIFF=$(git diff HEAD --stat 2>/dev/null | head -30 || echo "No changes")
+CURRENT_BRANCH=$(git branch --show-current 2>/dev/null || echo "unknown")
+```
+</step>
+
+<step name="fetch_issue_details">
+**Fetch full issue bodies from GitHub for matching context:**
+
+If project.json has issues, fetch their bodies for the agent to compare against:
+
+```bash
+ISSUE_BODIES=""
+if [ -n "$PROJECT_JSON" ]; then
+  ISSUE_NUMBERS=$(echo "$PROJECT_JSON" | python3 -c "
+import json, sys
+p = json.load(sys.stdin)
+idx = p['current_milestone'] - 1
+if idx < len(p['milestones']):
+    m = p['milestones'][idx]
+    for issue in m.get('issues', []):
+        print(issue.get('github_number', ''))
+" 2>/dev/null)
+
+  for NUM in $ISSUE_NUMBERS; do
+    if [ -n "$NUM" ]; then
+      BODY=$(gh issue view "$NUM" --json number,title,body -q '"\(.number)|\(.title)|\(.body)"' 2>/dev/null || echo "")
+      if [ -n "$BODY" ]; then
+        ISSUE_BODIES="${ISSUE_BODIES}${BODY}\n---\n"
+      fi
+    fi
+  done
+fi
+```
+</step>
+
+<step name="spawn_classification_agent">
+**Spawn general-purpose agent for question classification:**
+
+```
+Task(
+  prompt="
+<files_to_read>
+- ./CLAUDE.md (Project instructions — if exists, follow all guidelines)
+- .agents/skills/ (Project skills — if dir exists, list skills, read SKILL.md for each, follow relevant rules)
+</files_to_read>
+
+You are a question classification agent for the MGW pipeline. Your job is to
+classify a question or observation against the current project context and
+recommend an action.
+
+<question>
+${QUESTION}
+</question>
+
+<current_milestone>
+${MILESTONE_CONTEXT}
+</current_milestone>
+
+<milestone_issues>
+${ALL_ISSUES_CONTEXT}
+</milestone_issues>
+
+<issue_bodies>
+${ISSUE_BODIES}
+</issue_bodies>
+
+<active_work>
+Current branch: ${CURRENT_BRANCH}
+Active issues in .mgw/:
+${ACTIVE_STATE}
+</active_work>
+
+<recent_changes>
+${RECENT_DIFF}
+</recent_changes>
+
+<classification_rules>
+
+Classify the question into exactly ONE of these categories:
+
+| Category | Criteria | Action |
+|----------|----------|--------|
+| In-scope | Directly relates to the current active issue being worked on | Include in current work — no new issue needed |
+| Adjacent | Relates to a DIFFERENT issue in the same milestone | Note it on that issue — suggest posting a comment |
+| Separate | Doesn't match any open issue in the milestone | Suggest filing a new issue with a title |
+| Duplicate | Matches an existing issue (same root cause or fix) | Point to the existing issue |
+| Out-of-scope | Beyond the current milestone entirely | Note for future planning |
+
+Decision process:
+1. Read the question carefully
+2. Compare against each issue title, body, and scope in the milestone
+3. Check if it relates to current active work (branch, diff, active state)
+4. Look for keyword/concept overlap with existing issues
+5. If no match, determine if it fits the milestone's theme or is out-of-scope
+
+</classification_rules>
+
+<output_format>
+Return a structured classification report:
+
+## Classification: ${CATEGORY}
+
+### Analysis
+- What the question is about (1-2 sentences)
+- Why this classification was chosen (1-2 sentences)
+
+### Related Issue
+- Issue number and title (if adjacent or duplicate)
+- Or 'N/A — current work' (if in-scope)
+- Or 'No matching issue' (if separate or out-of-scope)
+
+### Recommendation
+- Specific actionable next step
+- If 'separate': suggest an issue title and brief body
+- If 'adjacent': suggest a comment to post on the related issue
+- If 'in-scope': note what to include in current work
+- If 'duplicate': point to the matching issue
+- If 'out-of-scope': note for future milestone planning
+</output_format>
+",
+  subagent_type="general-purpose",
+  description="Classify question: ${QUESTION}"
+)
+```
+</step>
+
+<step name="present_result">
+**Present the classification result to the user:**
+
+Display the agent's report, then present the MGW action banner:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ MGW ► QUESTION ROUTING
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+${classification_report_from_agent}
+
+───────────────────────────────────────────────────────
+```
+</step>
+
+<step name="offer_actions">
+**Offer follow-up actions based on classification:**
+
+**If "Separate":**
+```
+AskUserQuestion(
+  header: "File New Issue?",
+  question: "Create a new issue from this observation?",
+  options: [
+    { label: "Yes", description: "File the suggested issue via /mgw:issue" },
+    { label: "No", description: "Note it and continue current work" }
+  ]
+)
+```
+
+If user says "Yes":
+```
+Suggested issue ready. To file it:
+
+  gh issue create --title "${suggested_title}" --body "${suggested_body}"
+
+Or use /mgw:project to add it to a future milestone.
+```
+
+**If "Adjacent":**
+```
+AskUserQuestion(
+  header: "Post Comment?",
+  question: "Post a note on the related issue #${related_number}?",
+  options: [
+    { label: "Yes", description: "Post observation as a comment on #${related_number}" },
+    { label: "No", description: "Note it and continue current work" }
+  ]
+)
+```
+
+If user says "Yes":
+```bash
+gh issue comment ${related_number} --body "> **MGW** · \`observation\` · $(node ~/.claude/get-shit-done/bin/gsd-tools.cjs current-timestamp --raw)
+
+Observation noted during work on #${ACTIVE_ISSUE_NUMBER}:
+
+${QUESTION}"
+```
+
+Report: "Comment posted on #${related_number}."
+
+**If "In-scope":**
+```
+This relates to your current work on #${ACTIVE_ISSUE_NUMBER}.
+No action needed — include it in your current implementation.
+```
+
+**If "Duplicate":**
+```
+This appears to duplicate #${duplicate_number} — ${duplicate_title}.
+No new issue needed. Check that issue for existing progress.
+```
+
+**If "Out-of-scope":**
+```
+This is beyond the current milestone scope.
+Consider adding it to a future milestone or filing it for backlog:
+
+  gh issue create --title "${suggested_title}" --label "backlog"
+```
+</step>
+
+</process>
+
+<success_criteria>
+- [ ] Question text parsed from $ARGUMENTS (or prompted)
+- [ ] project.json loaded for milestone + issue context
+- [ ] .mgw/active/ state loaded for current work context
+- [ ] Recent git diff gathered for change context
+- [ ] Issue bodies fetched from GitHub for comparison
+- [ ] Classification agent spawned with full context
+- [ ] Classification returned: in-scope, adjacent, separate, duplicate, or out-of-scope
+- [ ] Related issue identified (if adjacent or duplicate)
+- [ ] Actionable recommendation provided
+- [ ] Follow-up action offered (file issue, post comment, etc.)
+- [ ] Delegation boundary respected: agent reads code/state, MGW presents results
+</success_criteria>
+</output>
diff --git a/commands/help.md b/commands/help.md
index b15dbba..6d8f6e1 100644
--- a/commands/help.md
+++ b/commands/help.md
@@ -53,6 +53,10 @@ COMMANDS
   /mgw:pr [number] [--base b]  Create PR from GSD artifacts + issue context
   /mgw:link <ref> <ref>        Cross-reference issues/PRs/branches
 
+  Query
+  ─────
+  /mgw:ask <question>          Route a question — in-scope, adjacent, separate, duplicate, out-of-scope
+
   Maintenance
   ───────────
   /mgw:sync                    Reconcile .mgw/ state with GitHub