aidevops 2.52.1 → 2.53.2

package/.agent/workflows/plans.md

@@ -1,709 +0,0 @@
----
-description: Planning workflow with auto-complexity detection
-mode: subagent
-tools:
-  read: true
-  write: true
-  edit: true
-  bash: true
-  glob: true
-  grep: true
-  webfetch: true
-  task: true
----
-
-# Plans Workflow
-
-<!-- AI-CONTEXT-START -->
-
-## Quick Reference
-
-- **Purpose**: Save planning discussions as actionable tasks or plans
-- **Commands**: `/save-todo` (auto-detects), `/ready` (show unblocked), `/sync-beads` (sync to Beads)
-- **Principle**: Don't make user think about where to save
-
-**Files**:
-
-| File | Purpose |
-|------|---------|
-| `TODO.md` | All tasks (simple + plan references) with dependencies |
-| `todo/PLANS.md` | Complex execution plans with context |
-| `todo/tasks/prd-{name}.md` | Product requirement documents |
-| `todo/tasks/tasks-{name}.md` | Implementation task lists |
-| `.beads/` | Beads database (synced from TODO.md) |
-
-**Task ID Format**:
-
-| Pattern | Example | Meaning |
-|---------|---------|---------|
-| `tNNN` | `t001` | Top-level task |
-| `tNNN.N` | `t001.1` | Subtask |
-| `tNNN.N.N` | `t001.1.1` | Sub-subtask |
-
-**Dependency Syntax**:
-
-| Field | Example | Meaning |
-|-------|---------|---------|
-| `blocked-by:` | `blocked-by:t001,t002` | Cannot start until these are done |
-| `blocks:` | `blocks:t003` | Completing this unblocks these |
-| Indentation | 2 spaces | Parent-child relationship |
-
-**Workflow**:
-
-```text
-Planning Conversation → /save-todo → Auto-detect → Save appropriately
-                                                         ↓
-Future Session → "Work on X" → Load context → git-workflow.md
-                                                         ↓
-                              /ready → Show unblocked tasks
-                                                         ↓
-                              /sync-beads → Sync to Beads for graph view
-```
-
-<!-- AI-CONTEXT-END -->
-
-## Auto-Detection Logic
-
-When `/save-todo` is invoked, analyze the conversation for complexity signals:
-
-| Signal | Indicates | Action |
-|--------|-----------|--------|
-| Single action item | Simple | TODO.md only |
-| < 2 hour estimate | Simple | TODO.md only |
-| User says "quick" or "simple" | Simple | TODO.md only |
-| Multiple distinct steps | Complex | PLANS.md + TODO.md |
-| Research/design needed | Complex | PLANS.md + TODO.md |
-| > 2 hour estimate | Complex | PLANS.md + TODO.md |
-| Multi-session work | Complex | PLANS.md + TODO.md |
-| PRD mentioned or needed | Complex | PLANS.md + TODO.md + PRD |
-
-## Ralph Classification
-
-Tasks can be classified as "Ralph-able" - suitable for autonomous iterative AI loops.
-
-### Ralph Criteria
-
-A task is Ralph-able when it has:
-
-| Criterion | Required | Example |
-|-----------|----------|---------|
-| **Clear success criteria** | Yes | "All tests pass", "Zero linting errors" |
-| **Automated verification** | Yes | Tests, linters, type checkers |
-| **Bounded scope** | Yes | Single feature, specific bug fix |
-| **No human judgment needed** | Yes | No design decisions, no UX choices |
-| **Deterministic outcome** | Preferred | Same input → same expected output |
-
-### Ralph Signals in Conversation
-
-| Signal | Ralph-able? | Why |
-|--------|-------------|-----|
-| "Make all tests pass" | Yes | Clear, verifiable |
-| "Fix linting errors" | Yes | Automated verification |
-| "Implement feature X with tests" | Yes | Tests provide verification |
-| "Refactor until clean" | Maybe | Needs specific criteria |
-| "Make it look better" | No | Subjective, needs human judgment |
-| "Design the API" | No | Requires design decisions |
-| "Debug production issue" | No | Unpredictable, needs investigation |
-
-### Tagging Ralph-able Tasks
-
-When a task meets Ralph criteria, add the `#ralph` tag:
-
-```markdown
-- [ ] t042 Fix all ShellCheck violations in scripts/ #ralph ~2h
-- [ ] t043 Implement user auth with tests #ralph #feature ~4h
-- [ ] t044 Design new dashboard layout #feature ~3h  (NOT ralph-able)
-```
-
-### Ralph Task Requirements
-
-When tagging a task as `#ralph`, ensure it includes:
-
-1. **Completion promise**: What phrase signals success
-2. **Verification command**: How to check if done
-3. **Max iterations**: Safety limit (default: 20)
-
-**Full format:**
-
-```markdown
-- [ ] t042 Fix all ShellCheck violations #ralph ~2h
-  ralph-promise: "SHELLCHECK_CLEAN"
-  ralph-verify: "shellcheck .agent/scripts/*.sh"
-  ralph-max: 10
-```
-
-**Shorthand** (for simple cases):
-
-```markdown
-- [ ] t042 Fix all ShellCheck violations #ralph(SHELLCHECK_CLEAN) ~2h
-```
-
-### Running Ralph Tasks
-
-```bash
-# Start a Ralph loop for a tagged task
-/ralph-loop "$(grep 't042' TODO.md)" --completion-promise "SHELLCHECK_CLEAN" --max-iterations 10
-
-# Or use the task ID directly
-/ralph-task t042
-```
-
-### Ralph in PLANS.md
-
-For complex plans, mark Ralph-able phases:
-
-```markdown
-#### Progress
-
-- [ ] Phase 1: Research API endpoints ~1h
-- [ ] Phase 2: Implement core logic #ralph ~2h
-  ralph-promise: "ALL_TESTS_PASS"
-  ralph-verify: "npm test"
-- [ ] Phase 3: Design UI components ~2h (requires human review)
-- [ ] Phase 4: Integration tests #ralph ~1h
-  ralph-promise: "INTEGRATION_PASS"
-  ralph-verify: "npm run test:integration"
-```
-
-### Quality Loop Integration
-
-Built-in Ralph-able workflows:
-
-| Workflow | Command | Promise |
-|----------|---------|---------|
-| Preflight | `/preflight-loop` | `PREFLIGHT_PASS` |
-| PR Review | `/pr-loop` | `PR_APPROVED` |
-| Postflight | `/postflight-loop` | `RELEASE_HEALTHY` |
-
-These use `quality-loop-helper.sh` which applies Ralph patterns to quality workflows.
-
-## Saving Work
-
-### Step 1: Extract from Conversation
-
-- **Title**: Concise task/plan name
-- **Description**: What needs to be done
-- **Estimate**: Time estimate with breakdown `~Xh (ai:Xh test:Xh read:Xm)`
-- **Tags**: Relevant categories (#seo, #security, #feature, etc.)
-- **Context**: Key decisions, research findings, constraints discussed
-
-### Step 2: Present with Auto-Detection
-
-**For Simple tasks:**
-
-```text
-Saving to TODO.md: "{title}" ~{estimate}
-
-1. Confirm
-2. Add more details first
-3. Create full plan instead (PLANS.md)
-```
-
-**For Complex work:**
-
-```text
-This looks like complex work. Creating execution plan.
-
-Title: {title}
-Estimate: ~{estimate}
-Phases: {count} identified
-
-1. Confirm and create plan
-2. Simplify to TODO.md only
-3. Add more context first
-```
-
-### Step 3: Save Appropriately
-
-#### Simple Save (TODO.md only)
-
-Add to TODO.md Backlog:
-
-```markdown
-## Backlog
-
-- [ ] t{NNN} {title} #{tag} ~{estimate} logged:{YYYY-MM-DD}
-```
-
-**Format elements** (all optional except id and description):
-- `t{NNN}` - Unique task ID (auto-generated, never reused)
-- `@owner` - Who should work on this
-- `#tag` - Category (seo, security, browser, etc.)
-- `~estimate` - Time estimate with breakdown: `~4h (ai:2h test:1h read:30m)`
-- `logged:YYYY-MM-DD` - Auto-added when task created
-- `blocked-by:t001,t002` - Dependencies (cannot start until these done)
-- `blocks:t003` - What this unblocks when complete
-
-Respond:
-
-```text
-Saved: "{title}" to TODO.md (~{estimate})
-Start anytime with: "Let's work on {title}"
-```
-
-#### Complex Save (PLANS.md + TODO.md)
-
-1. **Create PLANS.md entry**:
-
-```markdown
-### [{YYYY-MM-DD}] {Title}
-
-**Status:** Planning
-**Estimate:** ~{estimate}
-**PRD:** [todo/tasks/prd-{slug}.md](tasks/prd-{slug}.md) (if needed)
-**Tasks:** [todo/tasks/tasks-{slug}.md](tasks/tasks-{slug}.md) (if needed)
-
-#### Purpose
-
-{Why this work matters - from conversation context}
-
-#### Progress
-
-- [ ] ({timestamp}) Phase 1: {description} ~{est}
-- [ ] ({timestamp}) Phase 2: {description} ~{est}
-
-#### Context from Discussion
-
-{Key decisions, research findings, constraints from conversation}
-
-#### Decision Log
-
-(To be populated during implementation)
-
-#### Surprises & Discoveries
-
-(To be populated during implementation)
-```
-
-2. **Add reference to TODO.md** (bidirectional linking):
-
-```markdown
-- [ ] {title} #plan → [todo/PLANS.md#{slug}] ~{estimate} logged:{YYYY-MM-DD}
-```
-
-3. **Optionally create PRD/tasks** if scope warrants (use `/create-prd`, `/generate-tasks`)
-
-Respond:
-
-```text
-Saved: "{title}"
-- Plan: todo/PLANS.md
-- Reference: TODO.md
-{- PRD: todo/tasks/prd-{slug}.md (if created)}
-{- Tasks: todo/tasks/tasks-{slug}.md (if created)}
-
-Start anytime with: "Let's work on {title}"
-```
-
-## Context Preservation
-
-Always capture from the conversation:
-- Decisions made and their rationale
-- Research findings
-- Constraints identified
-- Open questions
-- Related links or references mentioned
-
-This context goes into the PLANS.md entry under "Context from Discussion" so future sessions have full context.
-
-## Starting Work from Plans
-
-When user says "Let's work on X" or references a task/plan:
-
-### 1. Find Matching Work
-
-```bash
-grep -i "{keyword}" TODO.md
-grep -i "{keyword}" todo/PLANS.md
-ls todo/tasks/*{keyword}* 2>/dev/null
-```
-
-### 2. Load Context
-
-If PRD/tasks exist, read them for full context.
-
-### 3. Present with Auto-Selection
-
-```text
-Found: "{title}" (~{estimate})
-
-1. Start working (creates branch: {suggested-branch})
-2. View full details first
-3. Different task
-
-[Enter] or 1 to start
-```
-
-### 4. Follow git-workflow.md
-
-After branch creation, follow standard git workflow.
-
-## During Implementation
-
-### Update Progress
-
-After each work session, update `todo/PLANS.md`:
-
-```markdown
-#### Progress
-
-- [x] (2025-01-14 10:00Z) Research API endpoints
-- [x] (2025-01-14 14:00Z) Create MCP server skeleton
-- [ ] (2025-01-15 09:00Z) Implement core tools ← IN PROGRESS
-```
-
-### Record Decisions
-
-When making significant choices:
-
-```markdown
-#### Decision Log
-
-- **Decision:** Use TypeScript + Bun stack
-  **Rationale:** Matches existing MCP patterns, faster builds
-  **Date:** 2025-01-14
-```
-
-### Note Surprises
-
-When discovering unexpected information:
-
-```markdown
-#### Surprises & Discoveries
-
-- **Observation:** Ahrefs rate limits are per-minute, not per-day
-  **Evidence:** API docs state 500 requests/minute
-  **Impact:** Need to implement request queuing
-```
-
-### Check Off Tasks
-
-Update `todo/tasks/tasks-{slug}.md` as work completes:
-
-```markdown
-- [x] 1.1 Research API endpoints
-- [x] 1.2 Document authentication flow
-- [ ] 1.3 Implement auth handler ← CURRENT
-```
-
-## Completing a Plan
-
-### 1. Mark Tasks Complete
-
-Ensure all tasks in `todo/tasks/tasks-{slug}.md` are checked.
-
-### 2. Record Time
-
-At commit time, offer time tracking:
-
-```text
-Committing: "{title}"
-Session duration: 2h 12m
-Estimated: ~4h
-
-1. Accept 2h 12m as actual
-2. Enter different time
-3. Skip time tracking
-```
-
-### 3. Update PLANS.md Status
-
-```markdown
-**Status:** Completed
-
-#### Outcomes & Retrospective
-
-**What was delivered:**
-- {Deliverable 1}
-- {Deliverable 2}
-
-**Time Summary:**
-- Estimated: 4h
-- Actual: 3h 15m
-- Variance: -19%
-```
-
-### 4. Update TODO.md
-
-Mark the reference task done:
-
-```markdown
-## Done
-
-- [x] {title} #plan → [todo/PLANS.md#{slug}] ~4h actual:3h15m completed:2025-01-15
-```
-
-### 5. Update CHANGELOG.md
-
-Add entry following `workflows/changelog.md` format.
-
-## PRD and Task Generation
-
-For complex work that needs detailed planning:
-
-### Generate PRD (`/create-prd`)
-
-Ask clarifying questions using numbered options:
-
-```text
-To create the PRD, I need to clarify:
-
-1. What is the primary goal?
-   A. {option}
-   B. {option}
-
-2. Who is the target user?
-   A. {option}
-   B. {option}
-
-Reply with "1A, 2B" or provide details.
-```
-
-Create PRD in `todo/tasks/prd-{slug}.md` using `templates/prd-template.md`.
-
-### Generate Tasks (`/generate-tasks`)
-
-**Phase 1: Parent Tasks**
-
-```text
-High-level tasks with estimates:
-
-- [ ] 0.0 Create feature branch ~5m
-- [ ] 1.0 {First major task} ~2h
-- [ ] 2.0 {Second major task} ~3h
-
-Total: ~5h 5m
-
-Reply "Go" to generate sub-tasks.
-```
-
-**Phase 2: Sub-Tasks**
-
-```markdown
-- [ ] 0.0 Create feature branch
-  - [ ] 0.1 Create and checkout: `git checkout -b feature/{slug}`
-
-- [ ] 1.0 {First major task}
-  - [ ] 1.1 {Sub-task}
-  - [ ] 1.2 {Sub-task}
-```
-
-Create in `todo/tasks/tasks-{slug}.md` using `templates/tasks-template.md`.
-
-## Time Estimation Heuristics
-
-| Task Type | AI Time | Test Time | Read Time |
-|-----------|---------|-----------|-----------|
-| Simple fix | 15-30m | 10-15m | 5m |
-| New function | 30m-1h | 15-30m | 10m |
-| New component | 1-2h | 30m-1h | 15m |
-| New feature | 2-4h | 1-2h | 30m |
-| Architecture change | 4-8h | 2-4h | 1h |
-| Research/spike | 1-2h | - | 30m |
-
-## Dependencies and Blocking
-
-### Dependency Syntax
-
-Tasks can declare dependencies using these fields:
-
-```markdown
-- [ ] t001 Parent task ~4h
-  - [ ] t001.1 Subtask ~2h blocked-by:t002
-    - [ ] t001.1.1 Sub-subtask ~1h
-  - [ ] t001.2 Another subtask ~1h blocks:t003
-```
-
-| Field | Syntax | Meaning |
-|-------|--------|---------|
-| `blocked-by:` | `blocked-by:t001,t002` | Cannot start until t001 AND t002 are done |
-| `blocks:` | `blocks:t003,t004` | Completing this task unblocks t003 and t004 |
-| Indentation | 2 spaces per level | Implicit parent-child relationship |
-
-### TOON Dependencies Block
-
-Dependencies are also stored in machine-readable TOON format:
-
-```markdown
-<!--TOON:dependencies[N]{from_id,to_id,type}:
-t019.2,t019.1,blocked-by
-t019.3,t019.2,blocked-by
-t020,t019,blocked-by
--->
-```
-
-### /ready Command
-
-Show tasks with no open blockers (ready to work on):
-
-```bash
-# Invoked via AI assistant
-/ready
-
-# Or via script
-~/.aidevops/agents/scripts/todo-ready.sh
-```
-
-Output:
-
-```text
-Ready to work (no blockers):
-
-1. t011 Demote wordpress.md from main agent to subagent ~1h
-2. t014 Document RapidFuzz library ~30m
-3. t004 Add Ahrefs MCP server integration ~2d
-
-Blocked (waiting on dependencies):
-
-- t019.2 Phase 2: Bi-directional sync (blocked-by: t019.1)
-- t020 Git Issues Sync (blocked-by: t019)
-```
-
-### Hierarchical Task IDs
-
-Tasks use stable, hierarchical IDs that are never reused:
-
-| Level | Pattern | Example | Use Case |
-|-------|---------|---------|----------|
-| Top-level | `tNNN` | `t001` | Independent tasks |
-| Subtask | `tNNN.N` | `t001.1` | Phases, components |
-| Sub-subtask | `tNNN.N.N` | `t001.1.1` | Detailed steps |
-
-**Rules:**
-- IDs are assigned sequentially and never reused
-- Subtasks inherit parent's ID as prefix
-- Maximum depth: 3 levels (t001.1.1)
-- IDs are stable across syncs with Beads
-
-## Beads Integration
-
-### Sync with Beads
-
-aidevops Tasks & Plans syncs bi-directionally with [Beads](https://github.com/steveyegge/beads) for graph visualization and analytics.
-
-```bash
-# Sync TODO.md → Beads
-/sync-beads push
-
-# Sync Beads → TODO.md
-/sync-beads pull
-
-# Two-way sync with conflict detection
-/sync-beads
-
-# Or via script
-~/.aidevops/agents/scripts/beads-sync-helper.sh [push|pull|sync]
-```
-
-### Sync Guarantees
-
-| Guarantee | Implementation |
-|-----------|----------------|
-| No race conditions | Lock file during sync |
-| Data integrity | Checksum verification before/after |
-| Conflict detection | Warns if both sides changed |
-| Audit trail | All syncs logged to `.beads/sync.log` |
-| Command-led only | No automatic sync (user controls timing) |
-
-### Beads UIs
-
-After syncing, use Beads ecosystem for visualization:
-
-| UI | Command | Best For |
-|----|---------|----------|
-| beads_viewer | `bv` | Graph analytics, PageRank, critical path |
-| beads-ui | `npx beads-ui start` | Web dashboard, kanban |
-| bdui | `bdui` | Quick terminal view |
-| perles | `perles` | BQL queries |
-| beads.el | `M-x beads-list` | Emacs users |
-
-## Time Tracking Configuration
-
-Configure per-repo in `.aidevops.json`:
-
-```json
-{
-  "time_tracking": "prompt",
-  "features": ["planning", "time-tracking", "beads"]
-}
-```
-
-| Setting | Behavior |
-|---------|----------|
-| `true` | Always prompt for time at commit |
-| `false` | Never prompt (disable time tracking) |
-| `prompt` | Ask once per session, remember preference |
-
-Use `/log-time-spent` command to manually log time anytime.
-
-## Git Branch Strategy for TODO.md Changes
-
-TODO.md changes fall into two categories with different branch strategies:
-
-### Stay on Current Branch (Default)
-
-Most TODO.md changes should stay on the current branch:
-
-| Change Type | Example | Why Stay? |
-|-------------|---------|-----------|
-| Task discovered during work | "Found we need rate limiting while building auth" | Related context |
-| Subtask additions | Adding t019.2.1 while working on t019 | Must stay together |
-| Status updates | Moving task to In Progress, marking Done | Part of workflow |
-| Dependency updates | Adding `blocked-by:` when discovering blockers | Discovered in context |
-| Context notes | Adding notes to tasks you're actively working on | Preserves context |
-
-### Consider Dedicated Branch
-
-When adding **unrelated backlog items** (new ideas, tools to evaluate, future work):
-
-| Condition | Recommendation |
-|-----------|----------------|
-| No uncommitted changes | Offer branch choice |
-| Has uncommitted changes | Stay on current branch (lower friction) |
-| Adding 3+ unrelated items | Suggest batching on dedicated branch |
-
-**Prompt pattern** (when adding unrelated backlog items):
-
-```text
-Adding {N} backlog items unrelated to `{current-branch}`:
-- {item 1}
-- {item 2}
-
-1. Add to current branch (quick, may create PR noise)
-2. Create `chore/backlog-updates` branch (cleaner history)
-3. Add to main directly (TODO.md only, skip PR)
-```
-
-### Why Not Always Switch?
-
-A "always switch branches for TODO.md" rule fails the 80% universal applicability test:
-
-- ~45% of todo additions ARE related to current work
-- Branch switching adds 2-5 minutes overhead per switch
-- Uncommitted changes make switching complex (stash/pop)
-- Context is lost when separating related discoveries
-
-**Bottom line**: Use judgment. Related work stays together; unrelated backlog can optionally go to a dedicated branch.
-
-## Integration with Other Workflows
-
-| Workflow | Integration |
-|----------|-------------|
-| `git-workflow.md` | Branch names derived from tasks/plans |
-| `branch.md` | Task 0.0 creates branch |
-| `feature-development.md` | Auto-suggests planning for complex work |
-| `preflight.md` | Run before marking plan complete |
-| `changelog.md` | Update on plan completion |
-
-## Related
-
-- `feature-development.md` - Feature implementation patterns
-- `git-workflow.md` - Branch creation and management
-- `branch.md` - Branch naming conventions
-
-## Templates
-
-- `templates/prd-template.md` - PRD structure
-- `templates/tasks-template.md` - Task list format
-- `templates/todo-template.md` - TODO.md for new repos
-- `templates/plans-template.md` - PLANS.md for new repos