skillweave-lifecycle

facade	true
experimental	true
name	skillweave-lifecycle
description	Navigate SkillWeave lifecycles, profiles, planning state, testing gates, and next-action recommendations.
argument-hint	command="[status|recommend|switch|phases|plan|test]" bundle="[id]"

/skillweave-lifecycle

Canonical metadata is English. User-facing artifacts follow the output language setting.

Facade and router for the SkillWeave lifecycle.
This skill navigates phases, recommends bundles, manages the planning board, and routes work to specialized subskills. It does NOT own execution engine behavior.

Delegation Model

Concern	Delegated To
Execution (Ralph Loop, lanes, batches)	skillweave-promptchain-execute
Observation (reports, timing, events)	skillweave-observe
Release (version, package, sign, publish)	skillweave-releasechain
Launch (deploy, communicate, go-live)	skillweave-launch

Mandatory Pre-Flight: SkillWeave Sandboxing

Before generating any output, verify and enforce the SkillWeave sandbox:

1. Enforce .skillweave/ Directory Structure

If .skillweave/ does not exist, create it:

.skillweave/
.skillweave/tracking-log/
.skillweave/templates/
.skillweave/sequences/
.skillweave/phases/
.skillweave/bundles/

2. Route All Outputs Into .skillweave/

All lifecycle reports, phase transitions, bundle recommendations, and navigation logs MUST be saved under .skillweave/phases/ or .skillweave/bundles/.

3. Git Isolation

Ensure .skillweave/ is listed in .gitignore.

4. Default Config

If .skillweave/config.yaml does not exist, create it with lifecycle defaults.

Proceed only after these four criteria are met.

Usage

Invoke the skill by its name with arguments. The skill is host-neutral; no executable prefix is required — route it through any host on any supported transport (Markdown or MCP).

# Navigator & Status
skillweave-lifecycle command="status"                    # Current phase + bundle + recommendation
skillweave-lifecycle command="recommend"                 # Next recommended bundle/action
skillweave-lifecycle command="phases"                    # List all 7 phases
skillweave-lifecycle command="switch" bundle="full-lifecycle"  # Switch to bundle

# Planning Board
skillweave-lifecycle command="plan"                      # Show planning board
skillweave-lifecycle command="plan" action="create" title="My Task" priority="high" type="feature"
skillweave-lifecycle command="plan" action="move" id="FEAT-001" target="doing"
skillweave-lifecycle command="plan" action="show" id="FEAT-001"
skillweave-lifecycle command="plan" action="seed" prd=".skillweave/prds/v1.0/prd.json"

# Testing
skillweave-lifecycle command="test"                      # Run all test levels
skillweave-lifecycle command="test" level="unit"         # Run specific level
skillweave-lifecycle command="test" action="results"     # Show latest results

# Wizard & Reports
skillweave-lifecycle command="wizard"                    # 5-question guided entry (Layer 0)
skillweave-lifecycle command="report"                    # Show latest release report
skillweave-lifecycle command="report" action="generate"  # Generate report from current state
skillweave-lifecycle command="report" action="list"      # List all reports

Commands

command="plan"

File-based planning system using the beans-pattern (Markdown tickets with YAML frontmatter in directory-as-state). Provides Kanban-style workflow tracking without external tools.

Actions

Action	Description	Parameters
(none)	Show ASCII board summary	—
create	Create new ticket	title, priority, type, depends_on
move	Transition ticket between states	id, target (backlog/doing/done)
show	Show single ticket detail	id
seed	Auto-create tickets from prd.json	prd (path to prd.json)
board	Regenerate BOARD.md	—

Directory Structure

.skillweave/planning/
├── BOARD.md           # Auto-generated overview (regenerated on changes)
├── backlog/           # Tickets not yet started
│   ├── FEAT-001.md
│   └── ...
├── doing/             # Tickets in progress
│   └── FEAT-002.md
└── done/              # Completed tickets (with completion timestamp)
    └── INFRA-001.md

Ticket Format

Each ticket is a Markdown file with YAML frontmatter:

---
id: FEAT-001
title: "Multi-Level Testing Flow"
status: backlog
priority: high          # high | medium | low
type: feature           # feature | bugfix | infrastructure | qa | documentation
created: 2026-05-17
completed:              # Set when moved to done/
week: 2                 # Target week
depends_on: [INFRA-001]
estimated_effort: 8     # Hours
---

Description of the task.

## Acceptance Criteria

- [ ] Criterion 1
- [ ] Criterion 2

ID Generation

New tickets receive sequential IDs based on type:

• Features: FEAT-NNN
• Infrastructure: INFRA-NNN
• Bugs: BUG-NNN
• QA: QA-NNN
• Documentation: DOC-NNN

The next available number is determined by scanning existing tickets across all states.

BOARD.md Auto-Generation

After any create, move, or seed action, BOARD.md is regenerated with:

• Summary counts (Backlog / Doing / Done)
• Tables for each state sorted by priority then ID
• Critical path visualization
• Week plan showing scheduled work

Seeding from prd.json

action="seed" reads a prd.json file and creates one ticket per task entry:

• Maps id, title, description, acceptanceCriteria, priority, type, dependsOn, estimatedEffort, week
• Skips tickets that already exist (by ID)
• Regenerates BOARD.md after seeding

Integration with ReleaseChain

When ReleaseChain starts a task, the corresponding ticket is moved to doing/. When ReleaseChain completes a task (gate = PROMOTE), the ticket moves to done/ with completed timestamp set.

---

command="test"

Multi-level testing engine with 5-level pyramid and 3-state gate decisions. Runs automatically during ReleaseChain verify steps or manually via /skillweave test.

Parameters

Parameter	Values	Default	Description
level	lint, unit, e2e_smoke, acceptance, evidence	(all)	Run specific level only
action	run, results	run	Show results instead of running
run	run-id	(latest)	Specific run to display

Test Levels (Pyramid)

Level	Trigger	Gate Weight
1. Lint/TypeCheck	After code generation	15%
2. Unit Tests	After code generation	30%
3. E2E Smoke	Before release gate	25%
4. Acceptance	Before release gate	20%
5. Evidence/Groundedness	After Council output	10%

Gate Decision (3-State)

Gate	Condition	Action
PROMOTE	All levels pass, coverage ≥ minimum	Proceed to next phase
HOLD	Minor regression, coverage below minimum, 1 evidence gap	Suggest fix, allow manual override
ROLLBACK	Unit tests fail, lint critical errors, E2E smoke fails, >50% criteria unmet	Block progression, require fix

ReleaseChain Integration

Testing integrates at two Ralph Loop points:

1. Verify step (after Implement): Runs Lint + Unit → gate decides Fix/Retry vs Integrate
2. Review Gate (before Advance): Runs E2E Smoke + Acceptance → gate decides release readiness

Results Storage

Results stored as JSON at .skillweave/testing/results/yyyy-mm-dd-run-N.json with full per-level details, gate decision, and reasoning.

Configuration

See .skillweave/testing/test-config.yaml for level enable/disable, gate weights, thresholds, and language-specific commands.

See references/testing-flow.md for complete documentation.

---

command="status"

Reports current project phase, active bundle (if any), entry/exit conditions status, and next recommended action. Reads .skillweave/phases/current.yaml and .skillweave/bundles/active.yaml.

command="phases"

Lists all 7 lifecycle phases with their entry/exit conditions, skills used, and capabilities. Validates current phase against its entry conditions.

command="recommend"

Analyzes project state and recommends a bundle based on:

• Current phase and completed milestones
• Entry requirements met vs. missing
• Estimated effort vs. remaining work
• Sequence types available

Returns: Bundle ID, Confidence Score (0.0–1.0), Begründung (reasoning), und nächster Schritt (next action).

command="switch" bundle="[id]"

Transitions the project into the specified bundle. Validates that entry requirements for the bundle's first phase are met. Creates an active bundle record at .skillweave/bundles/active.yaml and a phase tracking file at .skillweave/phases/current.yaml.

command="wizard"

5-question guided entry for non-technical users (Layer 0 / Progressive Disclosure).

Questions:

1. What do you want to do? (new idea / continue project / test / get feedback / release)
2. Context question based on answer (project name, domain, etc.)
3. Complexity hint (quick task / medium project / large project)
4. Preferred language (DE/EN)
5. Confirmation: "Soll ich [skill] starten?" (Shall I start [skill]?)

Routes to the correct Layer 2 skill with pre-filled parameters.

command="report"

Generate or view structured release reports.

Action	Description
(none)	Show latest report
generate	Generate report from current build state
list	List all reports in .skillweave/reports/

Report Naming Convention

All reports follow: yyyy-mm-dd-topic.md

Examples:

• 2026-05-17-v1.0-release.md
• 2026-05-17-council-multi-model-diagnosis.md
• 2026-05-20-sprint-retrospective.md

Report Generation

When action="generate", reads:

• Latest test results from .skillweave/testing/results/
• Planning board state from .skillweave/planning/
• Handover signals from .skillweave/handover/
• Git log since last report

Produces a report using the template at .skillweave/templates/release-report.md with sections: Executive Summary, Changes, Metrics, Key Decisions, Known Issues, Next Steps.

Bundle-Vorschlag (Recommendation Engine)

When command="recommend" is used, the skill evaluates all 5 bundles against the current project state:

Confidence-Score (0.0–1.0)

Calculated from weighted criteria:

• Entry Requirements Met (weight: 0.4) — percentage of satisfied entry conditions
• Sequence Type Compatibility (weight: 0.3) — match between available sequence types and bundle requirements
• Phase Continuity (weight: 0.2) — how naturally the bundle follows from the current phase
• Historical Fit (weight: 0.1) — pattern match from past .skillweave/tracking-log/ entries

Begründung

Each recommendation includes a plain-text explanation:

• Warum dieses Bundle zum aktuellen Status passt
• Welche Entry Conditions erfüllt/nicht erfüllt sind
• Welcher Sequence Type empfohlen wird
• Nächster konkreter Schritt

Example Output

Bundle: full-lifecycle
Confidence: 0.87
Begründung: Projekt befindet sich in Discovery (Phase 1/7) ohne aktives Bundle.
             Entry Conditions "Problem statement" erfüllt. Full-Lifecycle deckt
             alle verbleibenden Phasen ab. Empfohlener Sequence Type: mixed.
Nächster Schritt: Switch zu full-lifecycle, dann PromptChain aus PRD generieren.

Integration mit phases.yaml + bundles.yaml

The skill reads and writes structured data in YAML format:

.skillweave/phases/current.yaml

phase: discovery
started: 2026-04-01
entry_conditions:
  problem_statement: true
  stakeholder_identified: true
exit_conditions:
  validated_problem: false
  opportunity_assessment: false

.skillweave/bundles/active.yaml

bundle: full-lifecycle
phases: [discovery, blueprint, design, build, release, launch, post-release]
current_phase_index: 0
started: 2026-04-01

Sandbox-Preflight

Upon invocation, the skill MUST:

1. Check .skillweave/ structure exists (create if missing)
2. Check .gitignore for .skillweave/ exclusion
3. Check .skillweave/config.yaml for lifecycle defaults
4. Check .skillweave/phases/ and .skillweave/bundles/ exist
5. Load phases/current.yaml and bundles/active.yaml if present
6. Determine invocation mode based on command parameter

If any preflight step fails, abort and report the issue before proceeding.

Testing

The following scenarios should be verified:

• Bundle-Empfehlung für Discovery-Phase: Recommend a bundle when the project is in Discovery with no active bundle. Expected: full-lifecycle or discovery-to-blueprint with Confidence > 0.7.
• Bundle-Empfehlung für Build-Phase: Recommend a bundle when working code exists and phases indicate Build is active. Expected: release-and-launch or design-and-build.
• Alle 7 Phasen werden angezeigt: command="phases" lists all seven phases with entry/exit conditions.
• Confidence-Score wird berechnet: Every recommendation includes a numeric Confidence-Score (0.0–1.0) with Begründung.
• Plan: Board anzeigbar: command="plan" produces readable board summary with counts per state.
• Plan: Ticket erstellen: command="plan" action="create" title="Test" creates ticket with next available ID.
• Plan: Ticket verschieben: command="plan" action="move" id="FEAT-001" target="doing" moves file between directories and updates frontmatter status.
• Plan: Seed from PRD: command="plan" action="seed" prd="path/to/prd.json" creates all tickets and generates BOARD.md.
• Plan: BOARD.md regeneriert: After any mutation, BOARD.md reflects current state.

Companion Files

• references/phases-reference.md — Full documentation of all 7 lifecycle phases
• references/bundles-reference.md — Full documentation of all 5 bundles
• references/plan-commands.md — File-based planning system (beans-pattern) commands and integration
• references/testing-flow.md — Multi-level testing pyramid, gate logic, and ReleaseChain integration
• references/navigator-detection.md — Phase auto-detection algorithm and recommendation engine
• references/meta-commands.md — 7 meta-commands (Layer 1) with routing logic