aiblueprint-cli 1.4.99 → 1.4.100

package/agents-config/skills/apex/steps/step-00b-economy.md

@@ -1,247 +1,20 @@
 ---
 name: step-00b-economy
-description: Economy mode overrides - no subagents, direct tool usage to save tokens
-load_condition: economy_mode = true
+description: Apply a low-budget APEX policy without weakening safety, scope control, or minimum evidence.
 ---
 
-# Economy Mode Overrides
+# Low-budget policy
 
-**This file is ONLY loaded when `-e` or `--economy` flag is active.**
+`-e` sets `{budget_policy}` to `low`.
 
-These instructions OVERRIDE the default behavior in all steps to save tokens by avoiding subagent launches.
+- Keep analysis narrow: inspect the most relevant files and commands first.
+- Prefer the main agent for tightly coupled work.
+- Use at most one subagent at a time, and only when separate context avoids greater cost or provides independent review.
+- Prefer targeted checks before broad suites; run broader checks when risk or project rules require them.
+- Summarize large outputs into artifacts and keep only decisive evidence in context.
+- Use the harness-selected model unless a cheaper role-specific route is available and adequate.
+- Stop expanding exploration after the task contract is supported by sufficient evidence.
 
----
-
-<why_economy_mode>
-**Purpose:** Reduce token usage for users with limited Claude Code plans.
-
-**Trade-offs:**
-- ✅ Uses ~70% fewer tokens
-- ✅ Faster execution (no agent overhead)
-- ⚠️ Less thorough exploration
-- ⚠️ No parallel research
-- ⚠️ May miss some context
-
-**When to use:**
-- Limited monthly token budget
-- Simple, well-defined tasks
-- Familiar codebase
-- Quick fixes or small features
-</why_economy_mode>
-
----
-
-<override_rules>
-
-## CRITICAL: Apply These Overrides to ALL Steps
-
-**When `{economy_mode}` = true, these rules OVERRIDE the default instructions:**
-
----
-
-### Override 1: No Subagent Launches
-
-**DEFAULT behavior (when economy_mode = false):**
-```
-Launch parallel agents:
-- Agent 1: explore-codebase
-- Agent 2: explore-docs
-- Agent 3: websearch
-```
-
-**ECONOMY behavior (when economy_mode = true):**
-```
-Use direct tools instead:
-- Glob to find files
-- Grep to search content
-- Read to examine files
-- find-docs only if current technical docs are necessary
-- exa-search only if broader online research is necessary
-```
-
-**NEVER launch sub-agents in economy mode.**
-
----
-
-### Override 2: Direct Tool Usage Pattern
-
-Instead of launching exploration agents, use this pattern:
-
-```
-1. Glob to find relevant files:
-   - Glob: "**/*auth*" or "**/*{keyword}*"
-   - Glob: "src/**/*.ts" for specific areas
-
-2. Grep to find specific code:
-   - Grep: "function login" or "class Auth"
-   - Grep: pattern in specific directory
-
-3. Read to examine found files:
-   - Read the most relevant 3-5 files
-   - Focus on files matching the task
-
-4. Online research ONLY if:
-   - Library documentation needed
-   - Unknown API or pattern
-   - Limit to 1-2 searches max
-   - Use /Users/melvynx/.agents/skills/find-docs/SKILL.md for technical docs
-   - Use /Users/melvynx/.agents/skills/exa-search/SKILL.md for broader web research
-```
-
----
-
-### Override 3: Reduced Exploration Scope
-
-**DEFAULT:** Explore comprehensively, find all related code
-**ECONOMY:** Focus on most likely locations only
-
-```
-Economy exploration strategy:
-1. Start with obvious paths (src/auth/, src/api/, etc.)
-2. Search for exact keywords from task
-3. Read only files directly related to task
-4. Skip "nice to have" context
-5. Stop exploring when you have enough to proceed
-```
-
----
-
-### Override 4: Skip Optional Steps
-
-**In economy mode, skip or minimize:**
-- Extensive documentation research (use existing knowledge)
-- Web searches for common patterns (use best practices you know)
-- Multiple rounds of exploration (one round is enough)
-- Deep pattern analysis (quick scan is sufficient)
-
-**Always do:**
-- Find the files to modify
-- Understand existing patterns (quick read)
-- Identify dependencies
-- Create the plan
-
----
-
-### Override 5: Leaner Validation
-
-**DEFAULT:** Launch code-reviewer agent for adversarial review
-**ECONOMY:** Self-review without subagent
-
-```
-Economy validation:
-1. Run typecheck and lint (required)
-2. Run affected tests (required)
-3. Quick self-review checklist:
-   - [ ] No obvious bugs
-   - [ ] Follows existing patterns
-   - [ ] Error handling present
-   - [ ] No security issues
-4. Skip adversarial review agent
-```
-
----
-
-### Override 6: Test Mode Adjustments
-
-**If both `{economy_mode}` and `{test_mode}` are true:**
-
-```
-Economy test strategy:
-1. Analyze tests with Glob + Grep (not explore-codebase agent)
-2. Read 1-2 similar test files for patterns
-3. Create essential tests only:
-   - 1 happy path test
-   - 1 error case test
-   - Skip edge cases unless critical
-4. Run tests directly (no agent)
-```
-
-</override_rules>
-
----
-
-<step_specific_overrides>
-
-## Step-by-Step Economy Overrides
-
-### Step 01: Analyze (Economy)
-```
-INSTEAD OF: 3 parallel exploration agents
-DO:
-1. Glob "**/*{keyword}*" for task-related files
-2. Grep for specific patterns in src/
-3. Read top 3-5 most relevant files
-4. Skip online research unless stuck. If needed, use find-docs for technical docs and exa-search for broader web research.
-```
-
-### Step 02: Plan (Economy)
-```
-Same as default - planning doesn't use agents
-```
-
-### Step 03: Execute (Economy)
-```
-Same as default - execution doesn't use agents
-```
-
-### Step 04: Validate (Economy)
-```
-INSTEAD OF: Comprehensive validation
-DO:
-1. Run typecheck + lint
-2. Run only affected tests
-3. Quick manual review
-4. Skip coverage analysis
-```
-
-### Step 05: Examine (Economy)
-```
-INSTEAD OF: code-reviewer agent
-DO:
-1. Self-review with checklist:
-   - Security: no injection, no secrets
-   - Logic: handles errors, edge cases
-   - Quality: follows patterns, readable
-2. Note any concerns
-3. Move to resolve if issues found
-```
-
-### Step 07: Tests (Economy)
-```
-INSTEAD OF: Deep test analysis with agents
-DO:
-1. Glob for existing test files
-2. Read 1 similar test file
-3. Create minimal test coverage
-4. Focus on critical paths only
-```
-
-### Step 08: Run Tests (Economy)
-```
-Same as default - test running doesn't use agents
-```
-
-</step_specific_overrides>
-
----
-
-<economy_indicator>
-**When economy mode is active, start each step with:**
-
-```
-⚡ ECONOMY MODE - Using direct tools, no subagents
-```
-
-This reminds both Claude and the user that economy mode is active.
-</economy_indicator>
-
----
+Low budget never relaxes authority, secret handling, destructive-action boundaries, introduced-regression handling, or required proof.
 
-<success_metrics>
-Economy mode is successful when:
-- No sub-agent launches
-- Direct Glob/Grep/Read usage instead
-- Fewer than 3 WebSearch calls total
-- Implementation still correct and working
-- Tests pass (if test_mode enabled)
-</success_metrics>
+Set `{budget_policy}=low` and `{budget_applied}=true`, then return to `step-00-init.md`.

package/agents-config/skills/apex/steps/step-00b-interactive.md

@@ -1,170 +1,21 @@
 ---
 name: step-00b-interactive
-description: Interactively configure APEX workflow flags
-returns_to: step-00-init.md
+description: Configure APEX intent policies interactively when the user requests the menu.
 ---
 
-# Step 0b: Interactive Configuration
+# Interactive policy configuration
 
-## MANDATORY EXECUTION RULES (READ FIRST):
+Show the current policy values, then ask only about policies the user wants to change:
 
-- 🛑 NEVER skip the interactive menu
-- 🛑 NEVER assume user preferences
-- ✅ ALWAYS use AskUserQuestion for flag selection
-- ✅ ALWAYS update all flag variables before returning
-- 📋 YOU ARE A CONFIGURATOR, not an implementer
-- 💬 FOCUS on flag configuration only
-- 🚫 FORBIDDEN to start any workflow steps
+- interaction: `standard` or `low`;
+- review: `risk-based` or `adversarial`;
+- artifacts: `minimal` or `verbose`;
+- test authoring: `risk-based`, `on`, or `off`;
+- proof: `risk-based` or `runtime`;
+- budget: `low` or `standard`;
+- orchestration: `auto`, `prefer-parallel`, or `direct`;
+- delivery: branch and pull-request requests.
 
-## CONTEXT BOUNDARIES:
+Explain that implementation autonomy does not silently authorize unrelated Git, deployment, provider, or communication actions.
 
-- Variables available: All current flag values from step-00-init
-- This sub-step updates: All flag variables based on user selection
-- Return to step-00-init.md after completion
-
-## YOUR TASK:
-
-Present an interactive menu for the user to enable/disable workflow flags.
-
----
-
-## EXECUTION SEQUENCE:
-
-### 1. Display Current Configuration
-
-Show current flag values:
-```
-**Current APEX Configuration:**
-
-| Flag | Status | Description |
-|------|--------|-------------|
-| Auto (`-a`) | {auto_mode ? "✓ ON" : "✗ OFF"} | Skip confirmations |
-| Examine (`-x`) | {examine_mode ? "✓ ON" : "✗ OFF"} | Adversarial review |
-| Save (`-s`) | {save_mode ? "✓ ON" : "✗ OFF"} | Save outputs to files |
-| Test (`-t`) | {test_mode ? "✓ ON" : "✗ OFF"} | Include test steps |
-| Verify (`-v`) | {verify_mode ? "✓ ON" : "✗ OFF"} | Launch & verify feature |
-| Economy (`-e`) | {economy_mode ? "✓ ON" : "✗ OFF"} | No subagents |
-| Branch (`-b`) | {branch_mode ? "✓ ON" : "✗ OFF"} | Verify/create branch |
-| PR (`-pr`) | {pr_mode ? "✓ ON" : "✗ OFF"} | Create pull request |
-| Teams (`-m`) | {teams_mode ? "✓ ON" : "✗ OFF"} | Agent Teams parallel execution |
-```
-
-### 2. Ask for Flag Changes
-
-Use AskUserQuestion with multiSelect:
-```yaml
-questions:
-  - header: "Configure"
-    question: "Select flags to TOGGLE (currently shown flags will flip their state):"
-    options:
-      - label: "Auto mode"
-        description: "{auto_mode ? 'Disable' : 'Enable'} - skip confirmations"
-      - label: "Examine mode"
-        description: "{examine_mode ? 'Disable' : 'Enable'} - adversarial review at end"
-      - label: "Save mode"
-        description: "{save_mode ? 'Disable' : 'Enable'} - save outputs to .claude/output/"
-      - label: "Test mode"
-        description: "{test_mode ? 'Disable' : 'Enable'} - include test creation/runner"
-      - label: "Verify mode"
-        description: "{verify_mode ? 'Disable' : 'Enable'} - launch app and verify feature works"
-    multiSelect: true
-```
-
-### 3. Ask for Additional Flags
-
-Use AskUserQuestion with multiSelect:
-```yaml
-questions:
-  - header: "More"
-    question: "Select additional flags to TOGGLE:"
-    options:
-      - label: "Economy mode"
-        description: "{economy_mode ? 'Disable' : 'Enable'} - no subagents, save tokens"
-      - label: "Branch mode"
-        description: "{branch_mode ? 'Disable' : 'Enable'} - verify/create git branch"
-      - label: "PR mode"
-        description: "{pr_mode ? 'Disable' : 'Enable'} - create pull request at end"
-      - label: "Teams mode"
-        description: "{teams_mode ? 'Disable' : 'Enable'} - Agent Teams parallel execution"
-      - label: "Done - keep current"
-        description: "No more changes, proceed with workflow"
-    multiSelect: true
-```
-
-### 4. Apply Changes
-
-For each selected flag, toggle its value:
-```
-IF "Auto mode" selected → {auto_mode} = !{auto_mode}
-IF "Examine mode" selected → {examine_mode} = !{examine_mode}
-IF "Save mode" selected → {save_mode} = !{save_mode}
-IF "Test mode" selected → {test_mode} = !{test_mode}
-IF "Verify mode" selected → {verify_mode} = !{verify_mode}
-IF "Economy mode" selected → {economy_mode} = !{economy_mode}
-IF "Branch mode" selected → {branch_mode} = !{branch_mode}
-IF "PR mode" selected → {pr_mode} = !{pr_mode}
-IF "Teams mode" selected → {teams_mode} = !{teams_mode}
-```
-
-**Special rules:**
-
-If Teams mode enabled, auto-enable tasks mode (teams requires task breakdown):
-```
-IF {teams_mode} = true → {tasks_mode} = true
-```
-
-If PR mode enabled, auto-enable branch mode:
-```
-IF {pr_mode} = true → {branch_mode} = true
-```
-
-### 5. Show Final Configuration
-
-Display updated configuration:
-```
-**Updated APEX Configuration:**
-
-| Flag | Status |
-|------|--------|
-| Auto | {auto_mode ? "✓ ON" : "✗ OFF"} |
-| Examine | {examine_mode ? "✓ ON" : "✗ OFF"} |
-| Save | {save_mode ? "✓ ON" : "✗ OFF"} |
-| Test | {test_mode ? "✓ ON" : "✗ OFF"} |
-| Verify | {verify_mode ? "✓ ON" : "✗ OFF"} |
-| Economy | {economy_mode ? "✓ ON" : "✗ OFF"} |
-| Branch | {branch_mode ? "✓ ON" : "✗ OFF"} |
-| PR | {pr_mode ? "✓ ON" : "✗ OFF"} |
-| Teams | {teams_mode ? "✓ ON" : "✗ OFF"} |
-```
-
-### 6. Return
-
-→ Return to step-00-init.md with all flags updated
-
----
-
-## SUCCESS METRICS:
-
-✅ Current configuration displayed
-✅ User able to toggle any flag
-✅ All selected flags properly toggled
-✅ PR mode auto-enables branch mode
-✅ Final configuration shown
-
-## FAILURE MODES:
-
-❌ Not showing current flag states
-❌ Forgetting to toggle selected flags
-❌ Not enabling branch_mode when pr_mode selected
-❌ Starting workflow instead of returning
-❌ **CRITICAL**: Using plain text prompts instead of AskUserQuestion
-
----
-
-## RETURN:
-
-After configuration complete, return to `./step-00-init.md` to continue initialization.
-
-<critical>
-Remember: This sub-step ONLY handles flag configuration. Return immediately after updating flags.
-</critical>
+Apply the selected policies, set `{interactive_applied}=true`, and return to `step-00-init.md` for the remaining policy routes.

package/agents-config/skills/apex/steps/step-00b-save.md

@@ -1,125 +1,24 @@
 ---
 name: step-00b-save
-description: Setup save output structure for APEX workflow
-returns_to: step-00-init.md
+description: Configure APEX artifact detail while preserving the always-on minimal run record.
 ---
 
-# Step 0b: Save Mode Setup
+# Artifact policy
 
-## MANDATORY EXECUTION RULES (READ FIRST):
+Minimal machine-readable state is always written under `.agents/apex/runs/<run-id>/`.
 
-- 🛑 NEVER start analysis or implementation
-- ✅ ALWAYS run the setup script to create output structure
-- ✅ ALWAYS capture the generated `{task_id}` from the script output
-- 📋 YOU ARE A SETUP MANAGER, not an implementer
-- 🚫 FORBIDDEN to start any analysis work
+With `-s`, set `{artifact_policy}=verbose` and additionally preserve useful human-readable analysis, plans, command summaries, review findings, and proof galleries in `{run_dir}/artifacts/`.
 
-## CONTEXT BOUNDARIES:
+Without `-s`, keep `{artifact_policy}=minimal`: write only the run state, event log, checkpoints, task graph, evidence index, and artifacts required to support claims.
 
-- Variables available: `{feature_name}`, `{task_description}`, all flag variables, `{branch_name}`
-- This sub-step sets: `{task_id}`, `{output_dir}`
-- Return to step-00-init.md after completion
+For every artifact, record:
 
-## YOUR TASK:
+- stable ID and type;
+- producing phase or task;
+- repository revision and environment;
+- timestamp and relative path;
+- whether a later change invalidated it.
 
-Create the output directory structure and initialize all output files for the APEX workflow.
+Never copy secrets, credentials, full environment files, or unnecessary personal data into artifacts.
 
----
-
-## OUTPUT STRUCTURE:
-
-When save_mode is enabled, all outputs go to the PROJECT directory:
-
-```
-.claude/output/apex/{task-id}/
-├── 00-context.md        # Params, user request, timestamp
-├── 01-analyze.md        # Analysis findings
-├── 02-plan.md           # Implementation plan
-├── 03-execute.md        # Execution log
-├── 04-validate.md       # Validation results
-├── 02b-tasks/           # Task breakdown (if -k or -m)
-│   ├── README.md
-│   └── task-NN-*.md
-├── 05-examine.md        # Review findings (if -x)
-├── 06-resolve.md        # Resolution log (if -x)
-├── 07-tests.md          # Test analysis (if -t)
-├── 08-run-tests.md      # Test runner log (if -t)
-├── 10-verify.md         # Feature verification (if -v)
-└── 09-finish.md         # PR creation (if -pr)
-```
-
-## EXECUTION SEQUENCE:
-
-### 1. Run Template Setup Script
-
-```bash
-bash {skill_dir}/scripts/setup-templates.sh \
-  "{feature_name}" \
-  "{task_description}" \
-  "{auto_mode}" \
-  "{examine_mode}" \
-  "{save_mode}" \
-  "{test_mode}" \
-  "{economy_mode}" \
-  "{branch_mode}" \
-  "{pr_mode}" \
-  "{interactive_mode}" \
-  "{tasks_mode}" \
-  "{verify_mode}" \
-  "{branch_name}" \
-  "{original_input}"
-```
-
-**Note:** Pass `{feature_name}` (without number prefix), NOT `{task_id}`.
-
-The script:
-- Auto-generates `{task_id}` = `NN-{feature_name}` (next available number)
-- Creates `.claude/output/apex/{task_id}/` directory
-- Initializes `00-context.md` with configuration and progress table
-- Pre-creates all step files from templates
-- Only creates files for enabled steps (examine, tests, PR)
-- Outputs the generated `{task_id}`
-
-### 2. Capture Output Variables
-
-From the script output, set:
-- `{task_id}` = the generated ID (e.g., `01-add-auth-middleware`)
-- `{output_dir}` = `.claude/output/apex/{task_id}`
-
-### 3. Return
-
-→ Return to step-00-init.md with `{task_id}` and `{output_dir}` set
-
----
-
-## SAVE OUTPUT PATTERN (for all subsequent steps):
-
-Each step uses this pattern:
-
-1. `bash {skill_dir}/scripts/update-progress.sh "{task_id}" "{step_num}" "{step_name}" "in_progress"`
-2. Append findings/outputs to the pre-created step file
-3. `bash {skill_dir}/scripts/update-progress.sh "{task_id}" "{step_num}" "{step_name}" "complete"`
-
----
-
-## SUCCESS METRICS:
-
-✅ Output directory created
-✅ `{task_id}` generated and captured
-✅ `{output_dir}` set
-✅ All template files initialized
-
-## FAILURE MODES:
-
-❌ Not capturing `{task_id}` from script output
-❌ Starting analysis before returning
-
----
-
-## RETURN:
-
-After setup complete, return to `./step-00-init.md` to continue initialization.
-
-<critical>
-Remember: This sub-step ONLY creates the output structure. Return immediately after setting {task_id} and {output_dir}.
-</critical>
+Set `{artifact_applied}=true` and return to `step-00-init.md`.