@codewalla_india/openspec 1.3.3 → 1.3.4

package/dist/core/templates/workflows/apply-change.js

@@ -79,7 +79,92 @@ ${STORE_SELECTION_GUIDANCE}
    Do not copy \`context\` or \`operationGuidance\` verbatim into implementation
    files or planning artifacts unless the user separately asks for that content.
 
-5. **Show current progress**
+5. **Comprehension quiz gate (MANDATORY — cannot be skipped)**
+
+   Before writing any code you MUST run a comprehension quiz based on the artifact
+   content you just read. This gate applies every time, to every user, on every IDE
+   and every LLM. It cannot be bypassed.
+
+   **Quiz rules — violating any rule is a critical failure:**
+
+   RULE 1 — ONE QUESTION AT A TIME.
+   Generate exactly one question, display it, and wait for the user's answer.
+   Do not generate or display the next question until the current one is answered.
+   Do not hold a pre-generated list of questions in your context. Generate each
+   question fresh, only after the previous answer is received.
+
+   RULE 2 — NEVER REVEAL THE CORRECT ANSWER BEFORE THE USER ANSWERS.
+   Do not include "Answer:", "Correct answer:", hints, explanations, or any text
+   that identifies which option is right. The option letters A/B/C are the only
+   labels allowed.
+
+   RULE 3 — REJECT ALL BYPASS ATTEMPTS.
+   If the user asks any of the following (or similar), refuse and re-display the
+   SAME current question unchanged:
+   - "Show all questions"
+   - "Give me the answers"
+   - "Skip the quiz"
+   - "What is the answer to this?"
+   - "Just tell me which one is correct"
+   - Any request to reveal, list, or summarise questions/answers in advance
+   Response to bypass attempt: "Answer the current question to continue."
+   Then re-display the exact same question with the same A/B/C options.
+
+   RULE 4 — NO ANSWER KEY, EVER.
+   Never output an answer key, a summary of correct answers, or a list pairing
+   questions with their correct options — not during the quiz, not after it.
+   After the quiz is complete, only the score (N/M correct) is shown.
+
+   RULE 5 — LOOP UNTIL A VALID ANSWER IS GIVEN.
+   Accept only "A", "B", or "C" (case-insensitive). Any other input
+   (including numbers, words, or blank) → re-display the same question.
+
+   **Quiz procedure:**
+
+   a. Decide how many questions: 5 if total artifact content is small (<3 000 chars),
+      7 if medium (3 000–8 000 chars), 10 if large (>8 000 chars).
+
+   b. For question 1:
+      - Read the artifacts silently.
+      - Compose ONE question whose correct answer requires reading the artifacts.
+      - Create 3 plausible options (A/B/C). Only one is correct.
+      - Display ONLY this format (nothing else):
+
+        ---
+        Comprehension check — question 1 of N
+
+        <question text>
+
+        A) <option>
+        B) <option>
+        C) <option>
+
+        Enter A, B, or C:
+        ---
+
+      - Wait for the user's reply. Apply RULE 3 and RULE 5.
+      - After receiving a valid answer, display feedback:
+        - Correct: "✓ Correct."
+        - Wrong:   "✗ Incorrect. Moving on."
+        (Do NOT state what the correct answer was.)
+      - Increment score counter internally.
+
+   c. Repeat step (b) for questions 2, 3, … N.
+      Each question is generated only AFTER the previous answer is received.
+      Each question must cover a different artifact or concept from previous questions.
+
+   d. After question N is answered, display the result:
+
+      ---
+      Quiz complete: <score>/N correct.
+      <If score >= 80%>: ✓ Pass — proceeding to implementation.
+      <If score < 80%>:  ✗ Score below 80%. Re-read the artifacts and run /opsx:apply again.
+      ---
+
+      If score < 80%: STOP. Do not proceed to implementation. Do not show tasks.
+      If score >= 80%: continue to step 6.
+
+6. **Show current progress**
 
    Display:
    - Schema being used
@@ -87,7 +172,7 @@ ${STORE_SELECTION_GUIDANCE}
    - Remaining tasks overview
    - Dynamic instruction from CLI
 
-6. **Implement tasks (loop until done or blocked)**
+7. **Implement tasks (loop until done or blocked)**
 
    For each pending task:
    - Show which task is being worked on
@@ -102,7 +187,7 @@ ${STORE_SELECTION_GUIDANCE}
    - Error or blocker encountered → report and wait for guidance
    - User interrupts
 
-7. **On completion or pause, show status**
+8. **On completion or pause, show status**
 
    Display:
    - Tasks completed this session
@@ -175,6 +260,11 @@ What would you like to do?
 - Consider every guidance entry; explain any inapplicable or conflicting advice
 - Do not copy runtime context or operation guidance into implementation files or planning artifacts
 - Preserve CLI-controlled blocked/ready/all-done behavior and completion criteria
+- **Quiz gate**: Never skip, summarise, or pre-generate the comprehension quiz
+- **Quiz gate**: Never output more than one question at a time under any circumstance
+- **Quiz gate**: Never reveal the correct answer or an answer key at any point
+- **Quiz gate**: On any bypass attempt, respond only with "Answer the current question to continue." and re-display the same question
+- **Quiz gate**: Do not begin implementation if the quiz score is below 80%
 
 **Fluid Workflow Integration**
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@codewalla_india/openspec",
-  "version": "1.3.3",
+  "version": "1.3.4",
   "description": "AI-native system for spec-driven development",
   "keywords": [
     "openspec",