aiblueprint-cli 1.4.42 → 1.4.43

package/claude-code-config/skills/apex/steps/step-01-analyze.md

@@ -11,14 +11,11 @@ next_step: steps/step-02-plan.md
 - 🛑 NEVER plan or design solutions - that's step 2
 - 🛑 NEVER create todos or implementation tasks
 - 🛑 NEVER decide HOW to implement anything
-- 🛑 NEVER mark analyze as "complete" BEFORE writing findings to `{output_dir}/01-analyze.md` (if save_mode)
 - ✅ ALWAYS focus on discovering WHAT EXISTS
 - ✅ ALWAYS report findings with file paths and line numbers
-- ✅ IF save_mode: ALWAYS use Edit tool to append ALL findings to `{output_dir}/01-analyze.md` BEFORE proceeding
 - 📋 YOU ARE AN EXPLORER, not a planner
 - 💬 FOCUS on "What is here?" NOT "What should we build?"
 - 🚫 FORBIDDEN to suggest implementations or approaches
-- 🚫 FORBIDDEN to skip file saving when save_mode is true
 
 ## 🧠 SMART AGENT STRATEGY
 
@@ -33,10 +30,9 @@ Before exploring, THINK about what information you need and launch the RIGHT age
 ## EXECUTION PROTOCOLS:
 
 - 🎯 Launch parallel exploration agents (unless economy_mode)
-- 💾 IF save_mode: Use Edit tool to append findings to `{output_dir}/01-analyze.md` AFTER agents complete and BEFORE showing summary
+- 💾 Append findings to output file (if save_mode)
 - 📖 Document patterns with specific file:line references
 - 🚫 FORBIDDEN to proceed until context is complete
-- 🚫 FORBIDDEN to proceed to step-02 without saving findings first (if save_mode)
 
 ## CONTEXT BOUNDARIES:
 
@@ -218,14 +214,9 @@ Find common patterns and pitfalls.
 [Task: websearch - Stripe webhook security best practices]
 ```
 
-### 4. Synthesize Findings + Save to File
+### 4. Synthesize Findings
 
-<critical>
-**IF `{save_mode}` = true: You MUST write findings to the file BEFORE presenting the summary or proceeding.**
-This is the most commonly skipped step. DO NOT skip it.
-</critical>
-
-Combine all agent results into structured context using this format:
+Combine results into structured context:
 
 ```markdown
 ## Codebase Context
@@ -265,9 +256,7 @@ Combine all agent results into structured context using this format:
 - Use httpOnly cookies for tokens
 ```
 
-**IF `{save_mode}` = true (MANDATORY - DO THIS NOW):**
-
-Use the **Edit tool** to append the full synthesis to `{output_dir}/01-analyze.md`. Replace the placeholder text `_Findings will be appended here as exploration progresses..._` with all the structured findings above.
+**If `{save_mode}` = true:** Append synthesis to 01-analyze.md
 
 ### 5. Infer Acceptance Criteria
 
@@ -285,26 +274,9 @@ Based on "{task_description}" and existing patterns:
 _These will be refined in the planning step._
 ```
 
-**If `{save_mode}` = true:** Append acceptance criteria to `{output_dir}/01-analyze.md` using Edit tool.
+**If `{save_mode}` = true:** Update 00-context.md with acceptance criteria
 
-### 6. Complete Save Output (if save_mode)
-
-<critical>
-**This step MUST happen BEFORE presenting the summary to the user.**
-The file must already contain all findings before you proceed.
-</critical>
-
-**If `{save_mode}` = true:**
-
-1. Verify `{output_dir}/01-analyze.md` contains the full analysis (Read it to confirm)
-2. Update progress:
-
-```bash
-bash {skill_dir}/scripts/update-progress.sh "{task_id}" "01" "analyze" "complete"
-bash {skill_dir}/scripts/update-progress.sh "{task_id}" "02" "plan" "in_progress"
-```
-
-### 7. Present Context Summary
+### 6. Present Context Summary
 
 **Always (regardless of auto_mode):**
 
@@ -328,6 +300,17 @@ Present summary and proceed directly to planning:
 Do NOT ask for user confirmation here - always proceed directly to step-02-plan.
 </critical>
 
+### 7. Complete Save Output (if save_mode)
+
+**If `{save_mode}` = true:**
+
+Append summary to `{output_dir}/01-analyze.md` then:
+
+```bash
+bash {skill_dir}/scripts/update-progress.sh "{task_id}" "01" "analyze" "complete"
+bash {skill_dir}/scripts/update-progress.sh "{task_id}" "02" "plan" "in_progress"
+```
+
 ---
 
 ## SUCCESS METRICS:
@@ -346,7 +329,6 @@ Do NOT ask for user confirmation here - always proceed directly to step-02-plan.
 
 ## FAILURE MODES:
 
-❌ **CRITICAL**: Skipping file save when save_mode is true (proceeding to plan without writing to 01-analyze.md)
 ❌ Starting to plan or design (that's step 2!)
 ❌ Suggesting implementations or approaches
 ❌ Missing obvious related files
@@ -376,5 +358,4 @@ Always proceed directly to `./step-02-plan.md` after presenting context summary.
 <critical>
 Remember: Analysis is ONLY about "What exists?" - save all planning for step-02!
 Do NOT ask for confirmation - proceed directly!
-IF save_mode = true: The 01-analyze.md file MUST contain all findings BEFORE you load step-02-plan.md.
 </critical>

package/claude-code-config/skills/apex/steps/step-02-plan.md

@@ -11,23 +11,19 @@ next_step: steps/step-03-execute.md
 
 - 🛑 NEVER start implementing - that's step 3
 - 🛑 NEVER write or modify code in this step
-- 🛑 NEVER mark plan as "complete" BEFORE writing plan to `{output_dir}/02-plan.md` (if save_mode)
 - ✅ ALWAYS structure plan by FILE, not by feature
 - ✅ ALWAYS include specific line numbers from analysis
 - ✅ ALWAYS map acceptance criteria to file changes
-- ✅ ALWAYS create a TaskList using TaskCreate tool to track all planned file changes
-- ✅ IF save_mode: ALWAYS use Edit tool to write the full plan to `{output_dir}/02-plan.md` BEFORE showing summary
 - 📋 YOU ARE A PLANNER, not an implementer
 - 💬 FOCUS on "What changes need to be made where?"
-- 🚫 FORBIDDEN to use Edit/Write on source code files (only on output files if save_mode)
+- 🚫 FORBIDDEN to use Edit, Write, or Bash tools
 
 ## EXECUTION PROTOCOLS:
 
 - 🎯 ULTRA THINK before creating the plan
-- 💾 IF save_mode: Use Edit tool to write the full plan to `{output_dir}/02-plan.md` AFTER creating the plan and BEFORE presenting summary
+- 💾 Save plan to output file (if save_mode)
 - 📖 Reference patterns from step-01 analysis
 - 🚫 FORBIDDEN to proceed until user approves plan (unless auto_mode)
-- 🚫 FORBIDDEN to proceed to step-03 without saving plan first (if save_mode)
 
 ## CONTEXT BOUNDARIES:
 
@@ -163,64 +159,9 @@ questions:
 - Risk 1: [potential issue and mitigation]
 ```
 
-**IF `{save_mode}` = true (MANDATORY - DO THIS NOW):**
+**If `{save_mode}` = true:** Append full plan to 02-plan.md
 
-<critical>
-Use the **Edit tool** to write the FULL plan to `{output_dir}/02-plan.md`. Replace the placeholder content with the complete plan including all file changes, testing strategy, acceptance criteria mapping, and risks.
-</critical>
-
-### 5. Create Team (if teams_mode - BEFORE TaskList)
-
-<critical>
-If `{teams_mode}` = true, you MUST create the team BEFORE creating any tasks.
-TeamCreate resets the task list — so the team must exist first, then tasks are added to it.
-Do NOT spawn any agents yet — that happens in step-03-execute-teams.
-</critical>
-
-**If `{teams_mode}` = true:**
-
-```
-TeamCreate:
-  team_name: "apex-{feature_name}"
-  description: "APEX: {task_description}"
-```
-
-→ This creates the team and its empty task list.
-→ All subsequent TaskCreate calls will use this team's task list.
-→ Agents are spawned later in step-03-execute-teams.
-
-### 6. Create Task List (MANDATORY)
-
-<critical>
-You MUST create a TaskList using TaskCreate for every planned change.
-This is NOT optional - the task list tracks progress through execution.
-If teams_mode is enabled, the team was already created in step 5 — tasks go into the team's task list.
-</critical>
-
-**Create one task per file change from the plan:**
-
-For each file in the plan, call `TaskCreate` with:
-- **subject**: `{action} {filepath}` (e.g., "Add validateToken to src/auth/handler.ts")
-- **description**: The full details of what changes are needed for this file, including specific functions, patterns to follow, and line references
-- **activeForm**: Present continuous form (e.g., "Adding validateToken to handler.ts")
-
-**Then set up dependencies using `TaskUpdate`:**
-- If file B depends on file A (e.g., B imports from A), use `addBlockedBy` to mark the dependency
-- This ensures execution follows the correct order
-
-**Example:**
-```
-TaskCreate: "Create auth types in src/types/auth.ts"
-  → description: "Create AuthToken interface, ValidateResult type. Follow pattern from src/types/user.ts"
-  → activeForm: "Creating auth types"
-
-TaskCreate: "Add validateToken to src/auth/handler.ts"
-  → description: "Add validateToken function that returns ValidateResult. Handle expired token error case."
-  → activeForm: "Adding validateToken to handler"
-  → Then: TaskUpdate with addBlockedBy: [auth-types-task-id]
-```
-
-### 7. Verify Plan Completeness
+### 5. Verify Plan Completeness
 
 Checklist:
 - [ ] All files identified - nothing missing
@@ -229,139 +170,11 @@ Checklist:
 - [ ] Test coverage - all paths have test strategy
 - [ ] In scope - no scope creep
 - [ ] AC mapped - every criterion has implementation
-- [ ] **TaskList created with all file changes**
-- [ ] **Task dependencies set correctly**
-
-### 8. Brainstorm Uncertainty Points
-
-<critical>
-Before proceeding, THINK about what you're NOT 100% certain about.
-DO NOT ask generic "is this plan good?" questions.
-Instead, identify SPECIFIC uncertainties and ask TARGETED questions.
-</critical>
-
-**ULTRA THINK: Identify Uncertainties**
-
-For each aspect of the plan, rate your confidence (High/Medium/Low):
-
-```markdown
-## Uncertainty Analysis
-
-| Aspect | Confidence | Uncertainty |
-|--------|------------|-------------|
-| File locations correct? | High/Medium/Low | [What's unclear] |
-| Patterns match codebase? | High/Medium/Low | [What's unclear] |
-| Dependencies complete? | High/Medium/Low | [What's unclear] |
-| Error handling approach? | High/Medium/Low | [What's unclear] |
-| User expectations for X? | High/Medium/Low | [What's unclear] |
-| Technical approach for Y? | High/Medium/Low | [What's unclear] |
-```
-
-**Identify TOP 1-4 uncertainties with Low or Medium confidence.**
 
-These are the ONLY things worth asking about.
+### 6. Present Plan for Approval
 
----
-
-### 9. Ask Smart Questions (if not auto_mode)
-
-**If `{auto_mode}` = true:**
-→ Skip questions, use your best judgment, proceed directly
-
-**If `{auto_mode}` = false AND you have uncertainties:**
-
-<critical>
-ONLY ask questions about things where:
-1. Multiple valid approaches exist AND the choice significantly impacts implementation
-2. User intent is genuinely ambiguous from the task description
-3. A wrong assumption could cause significant rework
-
-DO NOT ask about:
-- Implementation details you can decide yourself
-- Things you're "slightly unsure" about (just decide)
-- Generic plan approval (never ask "is this plan good?")
-</critical>
-
-**Question Types to Use:**
-
-**Type 1: Behavior Clarification**
-When user intent is genuinely unclear:
-```yaml
-questions:
-  - header: "Behavior"
-    question: "When [specific scenario], should the system [option A] or [option B]?"
-    options:
-      - label: "[Option A] (Recommended)"
-        description: "[What A does and why it might be preferred]"
-      - label: "[Option B]"
-        description: "[What B does and when it's better]"
-    multiSelect: false
-```
-
-**Type 2: Scope Clarification**
-When scope boundaries are unclear:
-```yaml
-questions:
-  - header: "Scope"
-    question: "Should we also [related thing] as part of this, or keep it focused on [core thing]?"
-    options:
-      - label: "Keep focused (Recommended)"
-        description: "Only do [core thing], faster implementation"
-      - label: "Include [related thing]"
-        description: "More complete but expands scope"
-    multiSelect: false
-```
-
-**Type 3: Technical Choice**
-When multiple valid technical approaches exist:
-```yaml
-questions:
-  - header: "Approach"
-    question: "For [specific technical decision], we can use [approach A] or [approach B]. Which fits your needs?"
-    options:
-      - label: "[Approach A] (Recommended)"
-        description: "[Tradeoffs of A - e.g., simpler but less flexible]"
-      - label: "[Approach B]"
-        description: "[Tradeoffs of B - e.g., more complex but extensible]"
-    multiSelect: false
 ```
-
-**Type 4: Edge Case Handling**
-When edge case behavior is unclear:
-```yaml
-questions:
-  - header: "Edge case"
-    question: "If [edge case scenario] occurs, should we [behavior A] or [behavior B]?"
-    options:
-      - label: "[Behavior A] (Recommended)"
-        description: "[What happens with A]"
-      - label: "[Behavior B]"
-        description: "[What happens with B]"
-    multiSelect: false
-```
-
-**If you have 0 uncertainties worth asking about:**
-→ Proceed directly to execution without asking anything
-
-**Example of GOOD questions:**
-- "When a user submits invalid email format, should we show inline validation or only validate on submit?"
-- "Should the loading state show a spinner on the button or a full-page skeleton?"
-- "For rate limiting, should we limit per user or per IP address?"
-
-**Example of BAD questions (NEVER ask these):**
-- ❌ "Does this plan look good to you?"
-- ❌ "Ready to proceed with implementation?"
-- ❌ "Should I use async/await or promises?" (just decide)
-- ❌ "Is the file structure correct?" (you already verified)
-
----
-
-### 10. Present Plan Summary & Approve
-
-**Always show the full plan summary (both auto and non-auto):**
-
-```
-**Implementation Plan Ready: {task_description}**
+**Implementation Plan Ready**
 
 **Overview:** [1 sentence summary]
 
@@ -369,146 +182,34 @@ questions:
 **New files:** {count} files
 **Tests:** {count} test files
 
-**Key decisions made:**
-- [Decision 1 from user responses or auto-decided]
-- [Decision 2 from user responses or auto-decided]
-
 **Estimated changes:**
 - `file1.ts` - Major changes (add function, handle errors)
 - `file2.ts` - Minor changes (imports, single call)
 - `file1.test.ts` - New test file
-
-**Acceptance Criteria Coverage:**
-- [ ] AC1: Satisfied by changes in `file1.ts`
-- [ ] AC2: Satisfied by changes in `file2.ts`
-
-**Risks & Considerations:**
-- [Risk 1 and mitigation]
 ```
 
----
-
 **If `{auto_mode}` = true:**
-→ Show the full summary above, then proceed directly to next step
+→ Skip confirmation, proceed directly to execution
 
 **If `{auto_mode}` = false:**
 
-<critical>
-Use EnterPlanMode / ExitPlanMode for plan approval — NOT AskUserQuestion.
-This gives the user proper plan review UI and supports "Execute and clear context".
-</critical>
-
-**Step 10a: Call `EnterPlanMode`**
-
-This transitions into plan mode so the user can review and approve.
-
-**Step 10b: Write the plan to the plan file**
-
-The system will provide a plan file path. Write the FULL plan to it, including APEX state for context survival.
-
-<critical>
-The plan file must contain ALL details from step 4 — do NOT strip sections.
-Include everything: Overview, Prerequisites, File Changes, Testing Strategy, Acceptance Criteria Mapping, Risks & Considerations, Key Decisions.
-</critical>
-
-```markdown
-# APEX Implementation Plan: {task_description}
-
-## APEX Workflow Context
-
-> **IMPORTANT:** This is an APEX workflow plan. If you are reading this after
-> "Execute and clear context", continue the APEX workflow by loading the next
-> step file listed below.
-
-### State Variables
-| Variable | Value |
-|----------|-------|
-| task_description | {task_description} |
-| task_id | {task_id} |
-| feature_name | {feature_name} |
-| auto_mode | {auto_mode} |
-| examine_mode | {examine_mode} |
-| save_mode | {save_mode} |
-| test_mode | {test_mode} |
-| economy_mode | {economy_mode} |
-| branch_mode | {branch_mode} |
-| pr_mode | {pr_mode} |
-| tasks_mode | {tasks_mode} |
-| teams_mode | {teams_mode} |
-| output_dir | {output_dir} |
-| branch_name | {branch_name} |
-
-### Next Step After Approval
-{if tasks_mode or teams_mode:}
-  Load `skills/workflow-apex/steps/step-02b-tasks.md`
-{else:}
-  Load `skills/workflow-apex/steps/step-03-execute.md`
-
----
-
-## Plan Overview
-[1-2 sentence high-level strategy and approach]
-
-## Prerequisites
-- [ ] Prerequisite 1 (if any)
-- [ ] Prerequisite 2 (if any)
-
-## File Changes
-
-[Full file-by-file plan from step 4 — every file with specific actions, line references, patterns to follow, edge cases]
-
-## Testing Strategy
-
-**New tests:**
-- [test files to create with coverage details]
-
-**Update existing:**
-- [existing test files to modify]
-
-## Acceptance Criteria Mapping
-- [ ] AC1: Satisfied by changes in `file1.ts`
-- [ ] AC2: Satisfied by changes in `file2.ts`
-
-## Key Decisions
-- [Decision 1 from user responses or auto-decided]
-- [Decision 2 from user responses or auto-decided]
-
-## Risks & Considerations
-- [Risk 1: potential issue and mitigation]
-- [Risk 2: potential issue and mitigation]
-```
-
-**Step 10c: Call `ExitPlanMode`**
-
-Include appropriate allowedPrompts for execution:
-
 ```yaml
-allowedPrompts:
-  - tool: Bash
-    prompt: "run build, typecheck, lint, test, and format commands"
-  - tool: Bash
-    prompt: "run git operations for branch management and commits"
-  - tool: Bash
-    prompt: "run APEX workflow scripts for progress tracking"
+questions:
+  - header: "Plan"
+    question: "Review the implementation plan. Ready to proceed?"
+    options:
+      - label: "Approve and execute (Recommended)"
+        description: "Plan looks good, start implementation"
+      - label: "Adjust plan"
+        description: "I want to modify specific parts"
+      - label: "Ask questions"
+        description: "I have questions about the plan"
+      - label: "Start over"
+        description: "Revise the entire plan"
+    multiSelect: false
 ```
 
-**Step 10d: Handle "Execute and clear context"**
-
-If the user chooses "Execute and clear context":
-- Context is cleared but the plan file survives
-- The agent re-reads the plan file on the next turn
-- The APEX Workflow Context section in the plan file tells the agent:
-  1. This is an APEX workflow — restore all state variables from the table
-  2. Load the next step file indicated in "Next Step After Approval"
-  3. Continue the workflow as normal
-
-<critical>
-The plan file MUST contain ALL state variables, the next step path, AND the full detailed plan.
-Without this, "Execute and clear context" breaks the APEX workflow.
-Never strip sections from the plan file — it must be the complete implementation reference.
-</critical>
-
-### 11. Complete Save Output (if save_mode)
+### 7. Complete Save Output (if save_mode)
 
 **If `{save_mode}` = true:**
 
@@ -519,7 +220,6 @@ Append to `{output_dir}/02-plan.md`:
 **Status:** ✓ Complete
 **Files planned:** {count}
 **Tests planned:** {count}
-**Uncertainties resolved:** {count}
 **Next:** step-03-execute.md
 **Timestamp:** {ISO timestamp}
 ```
@@ -532,30 +232,18 @@ Append to `{output_dir}/02-plan.md`:
 ✅ Logical dependency order established
 ✅ All acceptance criteria mapped to changes
 ✅ Test strategy defined
-✅ **Team created with TeamCreate BEFORE TaskList (if teams_mode)**
-✅ **TaskList created with TaskCreate for every file change**
-✅ **Task dependencies set with TaskUpdate (addBlockedBy)**
-✅ Uncertainty points identified and addressed
-✅ Smart, targeted questions asked (if not auto_mode and uncertainties exist)
-✅ **Plan approved via EnterPlanMode/ExitPlanMode (if not auto_mode)**
-✅ **Plan file contains APEX state variables for "Execute and clear context" survival**
+✅ User approved plan (or auto-approved)
 ✅ NO code written or modified
 ✅ Output saved (if save_mode)
 
 ## FAILURE MODES:
 
-❌ **CRITICAL**: Skipping file save when save_mode is true (proceeding without writing to 02-plan.md)
 ❌ Organizing by feature instead of file
 ❌ Vague actions like "add feature" or "fix issue"
 ❌ Missing test strategy
 ❌ Not mapping to acceptance criteria
 ❌ Starting to write code (that's step 3!)
-❌ **CRITICAL**: Not creating a TaskList with TaskCreate
-❌ **CRITICAL**: Asking generic "is this plan good?" questions
-❌ **CRITICAL**: Asking about implementation details you can decide yourself
-❌ **CRITICAL**: Not brainstorming uncertainties before asking questions
-❌ **CRITICAL**: Using AskUserQuestion for plan approval instead of EnterPlanMode/ExitPlanMode (when auto_mode=false)
-❌ **CRITICAL**: Plan file missing APEX state variables (breaks "Execute and clear context")
+❌ **CRITICAL**: Not using AskUserQuestion for approval
 
 ## PLANNING PROTOCOLS:
 
@@ -569,25 +257,8 @@ Append to `{output_dir}/02-plan.md`:
 
 ## NEXT STEP:
 
-**If `{auto_mode}` = true:** Proceed directly after plan summary.
-**If `{auto_mode}` = false:** Proceed after user approves via ExitPlanMode.
-
-**If "Execute and clear context" was chosen:**
-→ The agent re-reads the plan file, restores APEX state variables from the table, and loads the next step indicated in the plan file.
-
-**Routing:**
-
-**If `{tasks_mode}` = true OR `{teams_mode}` = true:**
-→ Load `./step-02b-tasks.md` to generate task breakdown with dependencies
-
-**If `{teams_mode}` = true (after task breakdown):**
-→ Flow continues to `./step-03-execute-teams.md` for Agent Team parallel execution
-
-**Otherwise:**
-→ Load `./step-03-execute.md` to start implementation
+After user approves via AskUserQuestion (or auto-proceed), load `./step-03-execute.md`
 
 <critical>
 Remember: Planning is ONLY about designing the approach - save all implementation for step-03!
-When auto_mode=false, ALWAYS use EnterPlanMode/ExitPlanMode for approval — never AskUserQuestion.
-The plan file MUST contain APEX state variables so "Execute and clear context" works.
 </critical>

package/claude-code-config/skills/apex/steps/step-03-execute.md

@@ -32,7 +32,6 @@ next_step: steps/step-04-validate.md
 - Files to modify are known from the plan
 - Patterns to follow are documented from step-01
 - Don't add features - stick to the plan
-- **If context was cleared ("Execute and clear context"):** The plan file contains all APEX state variables in the "APEX Workflow Context" section. Read the plan file first and restore all variables before proceeding.
 
 ## YOUR TASK: