Promptly

Developer prompt analytics -- capture, analyze, and review AI conversations tied to development work.

Quick Start

npm i -g @getpromptly/cli
promptly init            # Configure AI tools + install /track skill
promptly start           # Prompts for ticket ID interactively
# ... work with any supported AI coding tool ...
promptly finish          # Save session data
promptly serve           # Open local dashboard at localhost:3000
promptly report          # View stats (select time period)
promptly digest          # Weekly insights digest with trends

Native /track Commands

After promptly init, use the /track command directly in your AI tool:

Tool	Command
Claude Code	/track AUTH-123
Codex CLI	/track AUTH-123
Gemini CLI	/track AUTH-123
VS Code + Copilot	/track AUTH-123

All tools support: /track <ticket-id>, /track status, /track finish

Features

• Local-first: Everything runs on your machine. No account needed.
• Multi-tool support: Works with Claude Code, Gemini CLI, Codex CLI, Cursor, Windsurf, and VS Code/Copilot via MCP.
• Native /track command: Built-in slash commands for Claude Code, Codex CLI, Gemini CLI, and VS Code + Copilot.
• Auto-prompt: AI tools offer session tracking at conversation start — no manual setup needed.
• Session tracking: Tag conversations to tickets, track token usage and duration.
• Git-aware sessions: Automatically captures commits, branches, and diff stats made during each session.
• Auto-categorization: Sessions are classified as bug-fix, feature, refactor, investigation, testing, or docs based on ticket ID, git commits, and conversation content.
• Session intelligence: Quality scoring (1-5 stars), tool usage analytics (Bash, Read, Edit, etc.), subagent tracking, context window metrics, and prompt quality analysis -- computed automatically at finish time.
• Session replay: Step through conversation turns with timing, playback controls, and cumulative stats to see how a session unfolded.
• Analytics dashboard: Cost-per-project trends, parallel session detection, skill usage analytics, instruction file effectiveness tracking, and aggregate prompt quality.
• Built-in dashboard: View sessions, analytics, digest, and session replay in your browser at localhost:3000.
• Optional cloud sync: For teams that want a shared dashboard.

Deployment Tiers

	Free (Local)	Cloud
Install	npm i -g @getpromptly/cli	SaaS
Storage	SQLite on disk	Our infra
Dashboard	promptly serve (localhost)	app.getpromptly.xyz
Users	Single user	Multi-user, teams
Signup	None	Per-seat
Connectivity	Full offline	Zero ops

CLI Commands

Command	Description
promptly init	Configure MCP (interactive tool selector)
promptly start	Start logging (prompts for ticket ID)
promptly start <ticket-id>	Start logging for specific ticket
promptly finish	Finish session and save data
promptly status	Show current session status
promptly serve	Start local dashboard on localhost:3000
promptly report	Show stats (interactive period selector)
promptly digest	Weekly insights digest with trends
promptly login	Authenticate with cloud API
promptly teams	List teams with quick actions
promptly team set	Interactive team selector
promptly team set <slug>	Set default team by slug
promptly team unset	Clear default team
promptly skill install	Install /track command for all configured tools
promptly skill status	Check skill installation status

How It Works

Claude Code / Gemini CLI / Codex CLI
        |
        └──> MCP Server --> ~/.promptly/buffer.json --> SQLite
                                                          |
                                            promptly serve (localhost:3000)

1. promptly init auto-detects installed AI coding tools and registers the MCP server with each.
2. promptly start TICKET-123 creates a session and signals the MCP server to begin logging.
3. As you work with your AI coding tool, the MCP server captures all conversation turns to a local buffer.
4. promptly finish captures git activity (commits, branch, diff stats, instruction file changes), auto-categorizes the session, computes session intelligence (quality score, tool usage, subagent stats, context window metrics, prompt quality analysis), writes data to SQLite, and clears the buffer.
5. promptly serve serves a dashboard with sessions, analytics, digest, and session replay.

Data Storage

All data is stored locally at ~/.promptly/:

File	Purpose
config.json	CLI configuration (API URL, mode, token)
session.json	Active session state
buffer.json	MCP conversation buffer (cleared on finish)
promptly.db	SQLite database with all completed sessions

Cloud Mode

For teams that want a shared dashboard, sign up at getpromptly.xyz.

promptly login             # Opens browser to sign in
promptly teams             # List your teams
promptly team set          # Interactive team selector (or: promptly team set my-team)
promptly start TICKET-1    # Creates session in your team
promptly finish            # Saves to SQLite AND uploads to API

Team Features

• Create teams at app.getpromptly.xyz/teams
• Invite members via shareable link (Owner/Admin can invite)
• Role-based access: Owner, Admin, Member
• Shared sessions: All team members can view sessions created within the team
• Team analytics: View usage by developer, model, and tool
• Per-tool comparison: See which AI coding tools your team uses most (Claude Code, Cursor, Gemini CLI, etc.)
• Data export: Download sessions and analytics as CSV or JSON from the dashboard

See docs/SETUP.md for cloud setup details.

Development

This is a TypeScript monorepo using Turborepo and pnpm.

pnpm install
pnpm build          # Build all packages
pnpm dev            # Dev mode (watch)

Packages

Package	Path	Description
@getpromptly/shared	packages/shared	Shared types and utilities
@getpromptly/cli	packages/cli	CLI tool
@getpromptly/mcp-server	packages/mcp-server	MCP server for conversation capture

Documentation

• Setup Guide -- Installation and configuration
• Architecture -- Technical design and data flow

License

MIT