Puzzle Iaa Analytics

puzzle-iaa-analytics

CRITICAL - This skill orchestrates a multi-step analysis pipeline. Read each step file in references/steps/ before executing. Do not skip steps or fabricate data. CRITICAL - Step 01 is a mandatory gate. All config values MUST be explicitly confirmed by the user via numbered text replies. data_inventory.json MUST NOT be written until 1.8 Summary Confirmation. Step 01 not confirmed → MUST NOT proceed to Steps 02-06. CRITICAL - Use ae-cli exclusively for all data operations. Read references/adapters/ae-cli.md for command patterns. Use ae-cli analysis adhoc run (unified) for all ad-hoc analysis — no separate builder step. Never guess command names, flags, JSON payloads, project_id, event names, property names, or parameter formats. CRITICAL - The references/config/project_mapping.md file is the sole bridge between business concepts and database fields. {{event.xxx}} and {{prop.xxx}} are business concept placeholders, NOT actual database field names. When switching projects, only modify this file. CRITICAL - SKILL.md and references body MUST NEVER reference references/adapters/. Tool knowledge lives only in the adapter layer and does not pollute methodology.

When to Use

Use puzzle-iaa-analytics for IAA puzzle/casual game post-launch data review:

• Post-launch version review: identify core issues in retention, monetization, and level experience
• Provide data-driven priority recommendations for the next version
• Applicable to pure IAA (ad-monetized) puzzle/casual games

If user intent is raw ad-hoc analysis, single-metric queries, or non-game analytics, do NOT use this skill — use ae-analysis directly.

Global Conventions

Working Directory

All intermediate files written to {{workspace_dir}} (default: /tmp/puzzle-iaa-analytics/). Do NOT rely on LLM memory between steps.

Status Codes

Degradation Strategy

• At least 4/7 analysis steps (Steps 02-06) must succeed to enter Step 07 (synthesis).
• Individual query failures do not block the overall flow — mark as "Data Missing" in final report.
• Step 01 failure → STEP_FATAL.

User Confirmation (Mandatory Gate)

Step 01 is the sole interaction window. All confirmations use text replies: you list numbered options, the user enters numbers.

• All config final values MUST be explicitly confirmed by user entering a number.
• Exact-match candidates must be reviewed; no auto-accept with "default configuration."
• Zero fuzzy-match results must be reported; user decides manual entry or explicit skip.
• data_inventory.json and project_mapping.md write-back wait for 1.8 Summary Confirmation.
• Step 01 not confirmed → MUST NOT proceed to Steps 02-06.
• Only mark skipped when user explicitly chooses. Do NOT fabricate user replies.

Parallelism

• Steps 02, 03, 04: mutually independent, execute fully in parallel.
• Steps 05, 06: depend on Step 02 results.
• Step 07: depends on all Steps 02-06.
• Sub-analyses within each step are also mutually independent — fire in one parallel batch.

Language Convention

• User-facing output: Chinese (中文).
• Confirmation options MUST be bilingual: format [N] English name — Chinese meaning.
• {{event.xxx}}, {{prop.xxx}} are business concept placeholders.
• Actual field names come from references/config/project_mapping.md.

Metric Reference Convention

Each analysis step references metric IDs from references/metric_definitions.md (e.g., "Execute §1 Retention Decay"). The six-element analysis unit template ensures metrics are independently verifiable mathematical formulas.

Architecture

puzzle-iaa-analytics/
├── SKILL.md                        ← Orchestrator (this file)
├── references/                     ← Platform-visible content
│   ├── adapters/ae-cli.md          ← ae-cli command patterns
│   ├── config/project_mapping.md   ← Project config (sole bridge)
│   ├── steps/                      ← Step-by-step execution
│   │   ├── 01_validate_data.md
│   │   ├── 02_core_diagnostics.md
│   │   ├── 03_level_deep_dive.md
│   │   ├── 04_user_path_funnel.md
│   │   ├── 05_user_segmentation.md
│   │   ├── 06_data_driven_intervals.md
│   │   ├── 07_synthesis_report.md
│   │   └── 08_publish.md
│   ├── templates/                  ← Output templates
│   │   ├── final_report_template.md
│   │   ├── key_metrics_table.md
│   │   ├── churn_analysis.md
│   │   ├── user_profile.md
│   │   └── recommendations.md
│   ├── metric_definitions.md       ← 17 analysis units (§1-§17)
│   ├── methodology.md              ← Business methodology
│   ├── drilldown_guide.md          ← User drilldown semantics
│   └── segmentation_guide.md       ← Segmentation semantics
└── scripts/
    └── parse_curve.py              ← Knee-point detection

Step Overview

┌─────────────────────────────────────────┐
│          01 Data Validation             │
└────────────────┬────────────────────────┘
                 │
     ┌───────────┼───────────┐
     ▼           ▼           ▼
┌─────────┐ ┌─────────┐ ┌─────────┐
│ 02 Core │ │ 03 Level│ │ 04 Path │
│ Diag    │ │ Deep    │ │ Funnel  │
└────┬────┘ └────┬────┘ └────┬────┘
     │           │            │
     └─────┬─────┘            │
           ▼                  │
     ┌───────────┐            │
     │ 05 Segmen │            │
     └─────┬─────┘            │
           ▼                  │
     ┌───────────┐            │
     │ 06 Interv │            │
     └─────┬─────┘            │
           │                  │
           └────┬─────────────┘
                ▼
         ┌───────────┐
         │07 Synthesis│
         └─────┬─────┘
               ▼
         ┌───────────┐
         │08 Publish │
         └───────────┘

Core Index

Metric Definitions → metric_definitions.md

Reference Documents

Output Templates

Tool Adapter

Start Execution

Read and execute references/steps/01_validate_data.md.