sparkey

Sparkey Reflect

AI coding advisor that analyzes your Claude Code sessions and generates personalized coaching insights. Runs 100% locally — your conversation data never leaves your machine.

Install

As a Claude Code plugin (recommended)

Requires Claude Code 1.0.33+ and Python 3.11+.

# Step 1: Add the marketplace source
/plugin marketplace add Sparkey-AI/sparkey-reflect

# Step 2: Install the plugin
/plugin install sparkey@sparkey-reflect

The plugin auto-installs the sparkey-reflect PyPI package on first run.

As a standalone CLI tool

pip install sparkey-reflect

Usage

Via Claude Code (recommended)

/sparkey:reflect              # Weekly report (last 7 days)
/sparkey:reflect daily        # Yesterday's sessions
/sparkey:reflect monthly      # Last 30 days
/sparkey:reflect full         # Full quarter (90 days)
/sparkey:reflect 14           # Custom day count
/sparkey:reflect deep-dive prompt_engineering   # Skill deep-dive
/sparkey:reflect update-rules # Improve your CLAUDE.md

Via CLI

sparkey-reflect analyze --tool claude-code --days 7
sparkey-reflect analyze --tool cursor --days 30 --format json
sparkey-reflect status
sparkey-reflect rules
sparkey-reflect trends --days 30
sparkey-reflect learning-path

Parameters

Parameter	Description
daily	Analyze yesterday's sessions
weekly (default)	Last 7 days
monthly	Last 30 days
full	Full quarter — 90 days
<number>	Custom number of days
deep-dive <skill>	In-depth analysis of one skill
update-rules	Improve CLAUDE.md based on insights

Deep-dive skills

Skill	What it analyzes
prompt_engineering	Prompt clarity, specificity, structure
context_management	How you manage context window and file references
tool_mastery	Usage of available tools and features
conversation_flow	Session structure and conversation patterns
session_habits	Session length, time-of-day, and workflow patterns
rule_file_quality	CLAUDE.md and instruction file effectiveness
outcome_quality	Success rate and task completion quality

Teams edition

Get version code + project management integrations, trend tracking, usage analytics, and industry benchmarks at sparkey.ai.

Methodology

Scoring grounded in industry-standard frameworks (DORA, SPACE, DevEx) and validated against published research from GitClear and METR.

DORA is a program of Google Cloud. GitClear and METR are independent organizations. Sparkey is not affiliated with or endorsed by any of these entities.

License

MIT