Use when you have a written implementation plan to execute in a separate session with review checkpoints
npx skills add miles-knowbl/orchestrator --skill "journey-tracer"
Install specific skill from multi-skill repository
# Description
Track execution journey through phases and skills with structured logging, state persistence, and timeline visualization for post-mortem analysis.
# SKILL.md
name: journey-tracer
description: "Track execution journey through phases and skills with structured logging, state persistence, and timeline visualization for post-mortem analysis."
phase: DOCUMENT
category: meta
version: "1.0.0"
depends_on: ["retrospective"]
tags: [meta, logging, tracing, state-machine, execution-log]
Journey Tracer
Track the execution journey with structured logging and state persistence.
When to Use
- During loop execution — Capture every phase transition, skill completion, and gate decision
- Post-mortem analysis — Understand what happened and why during a loop run
- Performance analysis — Identify bottlenecks in the engineering process
- When you say: "trace execution", "what happened in this loop", "execution timeline"
Required Deliverables
| Deliverable | Location | Condition |
|---|---|---|
| Journey log | memory/journeys/{executionId}.json |
Always |
| Timeline summary | Included in retrospective | Always |
Core Concept
Journey tracing answers: "What exactly happened during this loop execution, and in what order?"
Phase Start → Skill Invoked → Decision Made → Gate Evaluated → Phase Complete
↓ ↓ ↓ ↓ ↓
[logged] [logged] [logged] [logged] [logged]
Every significant event is captured with timestamp, context, and outcome.
Event Schema
interface JourneyEvent {
id: string;
timestamp: string; // ISO 8601
category: 'phase' | 'skill' | 'gate' | 'decision' | 'system';
action: string; // "started", "completed", "approved", "rejected"
subject: string; // Phase name, skill ID, gate ID
context: Record<string, unknown>; // Additional metadata
duration?: number; // Milliseconds (for completed events)
outcome?: 'success' | 'failure' | 'skipped';
}
Event Categories
| Category | Events | Example |
|---|---|---|
phase |
started, completed, skipped | Phase IMPLEMENT started |
skill |
invoked, completed, skipped, failed | Skill implement completed (success) |
gate |
evaluated, approved, rejected | Gate spec-gate approved by human |
decision |
made, deferred, overridden | Decision: use Drizzle over Prisma |
system |
paused, resumed, error, recovered | Execution paused by user |
Timeline Visualization
INIT ─────────── SCAFFOLD ──── IMPLEMENT ─────────────── TEST ──── ...
├─ requirements ├─ architect ├─ implement ├─ test-gen
│ (2 min) │ (5 min) │ (45 min) │ (10 min)
├─ spec ├─ scaffold │ │
│ (8 min) │ (3 min) │ │
├─ [spec-gate] ├─ [arch-gate]│ │
│ APPROVED │ APPROVED │ │
Checklist
- [ ] All phase transitions logged with timestamps
- [ ] All skill invocations logged with duration
- [ ] All gate decisions logged with reason
- [ ] Key decisions captured with context
- [ ] Journey file persisted to memory/journeys/
- [ ] Timeline summary available for retrospective
Relationship to Other Skills
| Skill | Relationship |
|---|---|
retrospective |
Consumes journey data for loop analysis |
calibration-tracker |
Uses journey timing data for calibration |
loop-controller |
Journey traces the controller's decisions |
memory-manager |
Journey data persisted via memory service |
Key Principles
Log everything. Better to have too much data than too little for post-mortem.
Structured events. Every log entry follows the same schema for queryability.
Timestamps are mandatory. Duration is derived from start/end pairs.
Context captures why. Not just what happened, but the reasoning behind decisions.
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.