Merge pull request #1297 from gsd-build/fix/discuss-phase-workflow-bypass-1292

fix: prevent discuss-phase from ignoring workflow instructions

commands/gsd/discuss-phase.md

@@ -46,50 +46,11 @@ Context files are resolved in-workflow using `init phase-op` and roadmap/state t
 DISCUSS_MODE=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" config-get workflow.discuss_mode 2>/dev/null || echo "discuss")
 ```
 
-If `DISCUSS_MODE` is `"assumptions"`: **Follow the discuss-phase-assumptions.md workflow instead of the steps below.** Skip all remaining steps in this process section.
+If `DISCUSS_MODE` is `"assumptions"`: Read and execute @~/.claude/get-shit-done/workflows/discuss-phase-assumptions.md end-to-end.
 
-If `DISCUSS_MODE` is `"discuss"` (or unset, or any other value): Continue with the steps below (current behavior).
+If `DISCUSS_MODE` is `"discuss"` (or unset, or any other value): Read and execute @~/.claude/get-shit-done/workflows/discuss-phase.md end-to-end.
 
----
-
-1. Validate phase number (error if missing or not in roadmap)
-2. Check if CONTEXT.md exists (offer update/view/skip if yes)
-3. **Load prior context** — Read PROJECT.md, REQUIREMENTS.md, STATE.md, and all prior CONTEXT.md files
-4. **Scout codebase** — Find reusable assets, patterns, and integration points
-5. **Analyze phase** — Check prior decisions, skip already-decided areas, generate remaining gray areas
-6. **Present gray areas** — Multi-select: which to discuss? Annotate with prior decisions + code context
-7. **Deep-dive each area** — 4 questions per area, code-informed options, Context7 for library choices
-8. **Write CONTEXT.md** — Sections match areas discussed + code_context section
-9. Offer next steps (research or plan)
-
-**CRITICAL: Scope guardrail**
-- Phase boundary from ROADMAP.md is FIXED
-- Discussion clarifies HOW to implement, not WHETHER to add more
-- If user suggests new capabilities: "That's its own phase. I'll note it for later."
-- Capture deferred ideas — don't lose them, don't act on them
-
-**Domain-aware gray areas:**
-Gray areas depend on what's being built. Analyze the phase goal:
-- Something users SEE → layout, density, interactions, states
-- Something users CALL → responses, errors, auth, versioning
-- Something users RUN → output format, flags, modes, error handling
-- Something users READ → structure, tone, depth, flow
-- Something being ORGANIZED → criteria, grouping, naming, exceptions
-
-Generate 3-4 **phase-specific** gray areas, not generic categories.
-
-**Probing depth:**
-- Ask 4 questions per area before checking
-- "More questions about [area], or move to next? (Remaining: [list unvisited areas])"
-- Show remaining unvisited areas so user knows what's still ahead
-- If more → ask 4 more, check again
-- After all areas → "Ready to create context?"
-
-**Do NOT ask about (Claude handles these):**
-- Technical implementation
-- Architecture choices
-- Performance concerns
-- Scope expansion
+**MANDATORY:** The execution_context files listed above ARE the instructions. Read the workflow file BEFORE taking any action. The objective and success_criteria sections in this command file are summaries — the workflow file contains the complete step-by-step process with all required behaviors, config checks, and interaction patterns. Do not improvise from the summary.
 </process>
 
 <success_criteria>

get-shit-done/bin/lib/config.cjs

@@ -18,6 +18,8 @@ const VALID_CONFIG_KEYS = new Set([
   'workflow.nyquist_validation', 'workflow.ui_phase', 'workflow.ui_safety_gate',
   'workflow.auto_advance', 'workflow.node_repair', 'workflow.node_repair_budget',
   'workflow.text_mode',
+  'workflow.research_before_questions',
+  'workflow.discuss_mode',
   'workflow._auto_chain_active',
   'git.branching_strategy', 'git.phase_branch_template', 'git.milestone_branch_template', 'git.quick_branch_template',
   'planning.commit_docs', 'planning.search_gitignored',
@@ -28,6 +30,8 @@ const CONFIG_KEY_SUGGESTIONS = {
   'workflow.nyquist_validation_enabled': 'workflow.nyquist_validation',
   'agents.nyquist_validation_enabled': 'workflow.nyquist_validation',
   'nyquist.validation_enabled': 'workflow.nyquist_validation',
+  'hooks.research_questions': 'workflow.research_before_questions',
+  'workflow.research_questions': 'workflow.research_before_questions',
 };
 
 function validateKnownConfigKeyPath(keyPath) {
@@ -108,6 +112,8 @@ function buildNewProjectConfig(userChoices) {
       ui_phase: true,
       ui_safety_gate: true,
       text_mode: false,
+      research_before_questions: false,
+      discuss_mode: 'discuss',
     },
     hooks: {
       context_warnings: true,

get-shit-done/templates/config.json

@@ -7,7 +7,8 @@
     "verifier": true,
     "auto_advance": false,
     "nyquist_validation": true,
-    "discuss_mode": "discuss"
+    "discuss_mode": "discuss",
+    "research_before_questions": false
   },
   "planning": {
     "commit_docs": true,

get-shit-done/workflows/discuss-phase.md

@@ -570,7 +570,7 @@ Track deferred ideas internally.
 
 For each selected area, conduct a focused discussion loop.
 
-**Research-before-questions mode:** Check if `research_questions` is enabled in config (from init context or `.planning/config.json`). When enabled, before presenting questions for each area:
+**Research-before-questions mode:** Check if `workflow.research_before_questions` is enabled in config (from init context or `.planning/config.json`). When enabled, before presenting questions for each area:
 1. Do a brief web search for best practices related to the area topic
 2. Summarize the top findings in 2-3 bullet points
 3. Present the research alongside the question so the user can make a more informed decision

get-shit-done/workflows/new-project.md

@@ -218,7 +218,7 @@ Ask inline (freeform, NOT AskUserQuestion):
 
 Wait for their response. This gives you the context needed to ask intelligent follow-up questions.
 
-**Research-before-questions mode:** Check if `research_questions` is enabled in `.planning/config.json` (or the config from init context). When enabled, before asking follow-up questions about a topic area:
+**Research-before-questions mode:** Check if `workflow.research_before_questions` is enabled in `.planning/config.json` (or the config from init context). When enabled, before asking follow-up questions about a topic area:
 
 1. Do a brief web search for best practices related to what the user described
 2. Mention key findings naturally as you ask questions (e.g., "Most projects like this use X — is that what you're thinking, or something different?")

get-shit-done/workflows/settings.md

@@ -163,7 +163,10 @@ Merge new settings into existing config.json:
     "auto_advance": true/false,
     "nyquist_validation": true/false,
     "ui_phase": true/false,
-    "ui_safety_gate": true/false
+    "ui_safety_gate": true/false,
+    "text_mode": true/false,
+    "research_before_questions": true/false,
+    "discuss_mode": "discuss" | "assumptions"
   },
   "git": {
     "branching_strategy": "none" | "phase" | "milestone",
@@ -171,11 +174,7 @@ Merge new settings into existing config.json:
   },
   "hooks": {
     "context_warnings": true/false,
-    "workflow_guard": true/false,
-    "research_questions": true/false
-  },
-  "workflow": {
-    "text_mode": true/false  // Use plain-text questions instead of TUI menus (for /rc remote sessions)
+    "workflow_guard": true/false
   }
 }
 ```

tests/config.test.cjs

@@ -477,6 +477,60 @@ describe('config-new-project command', () => {
   });
 });
 
+// ─── config-set (research_before_questions and discuss_mode) ──────────────────
+
+describe('config-set research_before_questions and discuss_mode', () => {
+  let tmpDir;
+
+  beforeEach(() => {
+    tmpDir = createTempProject();
+    runGsdTools('config-ensure-section', tmpDir);
+  });
+
+  afterEach(() => {
+    cleanup(tmpDir);
+  });
+
+  test('workflow.research_before_questions is a valid config key', () => {
+    const result = runGsdTools('config-set workflow.research_before_questions true', tmpDir);
+    assert.ok(result.success, `Command failed: ${result.error}`);
+
+    const config = readConfig(tmpDir);
+    assert.strictEqual(config.workflow.research_before_questions, true);
+  });
+
+  test('workflow.discuss_mode is a valid config key', () => {
+    const result = runGsdTools('config-set workflow.discuss_mode assumptions', tmpDir);
+    assert.ok(result.success, `Command failed: ${result.error}`);
+
+    const config = readConfig(tmpDir);
+    assert.strictEqual(config.workflow.discuss_mode, 'assumptions');
+  });
+
+  test('research_before_questions defaults to false in new configs', () => {
+    const config = readConfig(tmpDir);
+    assert.strictEqual(config.workflow.research_before_questions, false);
+  });
+
+  test('discuss_mode defaults to discuss in new configs', () => {
+    const config = readConfig(tmpDir);
+    assert.strictEqual(config.workflow.discuss_mode, 'discuss');
+  });
+
+  test('hooks.research_questions is rejected with suggestion', () => {
+    const result = runGsdTools('config-set hooks.research_questions true', tmpDir);
+    assert.strictEqual(result.success, false);
+    assert.ok(
+      result.error.includes('Unknown config key'),
+      `Expected "Unknown config key" in error: ${result.error}`
+    );
+    assert.ok(
+      result.error.includes('workflow.research_before_questions'),
+      `Expected suggestion for workflow.research_before_questions in error: ${result.error}`
+    );
+  });
+});
+
 // ─── config-set (additional coverage) ────────────────────────────────────────
 
 describe('config-set unknown key (no suggestion)', () => {

tests/discuss-mode.test.cjs

@@ -26,6 +26,41 @@ describe('workflow.discuss_mode config', () => {
     assert.ok(command.includes('workflow.discuss_mode'), 'should reference config key');
   });
 
+  test('discuss-phase command process block defers to workflow file (not inline instructions)', () => {
+    const command = fs.readFileSync(
+      path.join(__dirname, '..', 'commands', 'gsd', 'discuss-phase.md'), 'utf8'
+    );
+    // Extract the <process> block
+    const processMatch = command.match(/<process>([\s\S]*?)<\/process>/);
+    assert.ok(processMatch, 'should have a <process> block');
+    const processBlock = processMatch[1];
+
+    // The process block must explicitly tell the agent to read the workflow file
+    assert.ok(
+      processBlock.includes('Read and execute'),
+      'process block should direct agent to read and execute workflow file'
+    );
+    assert.ok(
+      processBlock.includes('MANDATORY'),
+      'process block should include MANDATORY instruction to read workflow files'
+    );
+
+    // The process block must NOT contain detailed step-by-step instructions
+    // that could substitute for the actual workflow file
+    assert.ok(
+      !processBlock.includes('Scout codebase'),
+      'process block should not contain detailed workflow steps (Scout codebase)'
+    );
+    assert.ok(
+      !processBlock.includes('Deep-dive each area'),
+      'process block should not contain detailed workflow steps (Deep-dive)'
+    );
+    assert.ok(
+      !processBlock.includes('Probing depth'),
+      'process block should not contain detailed workflow steps (Probing depth)'
+    );
+  });
+
   test('discuss-phase command argument-hint includes --text', () => {
     const command = fs.readFileSync(
       path.join(__dirname, '..', 'commands', 'gsd', 'discuss-phase.md'), 'utf8'