Skilled

Product Introduction

1. Definition: Skilled is a local-first, terminal-based analytics dashboard (TUI) for AI-assisted coding tools. It is a privacy-centric observability platform that parses local session history files from various AI coding assistants.
2. Core Value Proposition: Skilled provides developers with actionable, real-time insights into their AI tool usage patterns without compromising data privacy. It exists to quantify and visualize skill invocation frequency, trends, and project context across multiple AI coding agents, enabling data-driven optimization of developer workflows.

Main Features

1. Live 30 FPS Terminal Dashboard: Renders a high-performance, real-time user interface in the terminal using the @opentui/core framework. The dashboard includes bar charts with 8-level Unicode block elements for sub-character precision, a 16-week activity heatmap with a 5-level green intensity ramp, an hourly distribution histogram, and a recent activity feed, all updating dynamically.
2. Multi-Provider Local File Parsing: Automatically detects and reads session history from installed AI tools. It includes dedicated providers for Claude Code (~/.claude/history.jsonl), OpenCode, Codex, Grok, and Droid. Each provider extracts skill invocations (slash commands, tool calls) into a normalized data model containing skill name, timestamp, project path, session ID, and source tool.
3. CLI with JSON Output and Skill Audit Engine: Offers a command-line interface for scripting and automation. Commands like skilled list, skilled audit, and skilled detail output structured JSON. The audit engine applies heuristics to categorize skills: "Rising" (50%+ usage increase over 4 weeks), "Stale" (unused for 30+ days), "Heavy Hitters" (high frequency), and "One-Offs," providing a health report for skill management.

Problems Solved

1. Pain Point: Developers lack visibility into their interaction patterns with AI coding assistants, leading to inefficient tool use, forgotten capabilities, and an inability to measure the ROI of different AI tools or specific skills across projects.
2. Target Audience: The primary users are software engineers, DevOps professionals, and technical leads who utilize multiple AI coding tools (e.g., Claude Code, GitHub Copilot) daily and seek to optimize their workflow efficiency. Secondary users include engineering managers wanting to understand team tool usage trends (when aggregated ethically).
3. Use Cases: A developer wants to identify their most-used code review skills to create custom shortcuts. A team lead audits if newly introduced refactoring skills are being adopted. An individual wants to clean up unused or stale skills from their prompt history. A user compares the hourly distribution of their AI tool usage to optimize focus periods.

Unique Advantages

1. Differentiation: Unlike cloud-based analytics platforms or tool-specific dashboards, Skilled operates entirely offline with zero network calls, zero telemetry, and no account requirements. It aggregates data across competing AI tools (Claude, Codex, Grok) into a single pane of glass, a capability not offered by the vendors themselves.
2. Key Innovation: Its privacy-by-design architecture and use of a high-performance Rust indexer for parsing large local history files. The combination of a real-time TUI, a normalized cross-provider data model, and offline-first operation creates a unique category of personal developer intelligence tooling.

Frequently Asked Questions (FAQ)

1. Does Skilled send my coding data to the cloud? No, Skilled is a strictly local application. It performs zero network operations, has no telemetry, and all data processing occurs on your machine. It only reads existing local history files created by your AI coding tools.
2. Which AI coding tools does Skilled support? Skilled currently supports Claude Code, OpenCode, Codex, Grok, and Droid. It auto-detects these tools by checking for their standard local history file paths. Support for new tools can be added by creating a new file parser "provider."
3. How do I install Skilled on my system? You can install Skilled via a shell script (curl to install.sh), npm (npm install -g @avcodes/skilled), or pip (pip install skilled). The installation includes the main TypeScript application and, where possible, a pre-compiled Rust indexer for performance.
4. Can I use Skilled's data for automated reports? Yes, the Skilled CLI supports a --json flag on all major commands (e.g., skilled list --json). This allows you to pipe machine-readable skill usage data, filtered by source or project, into other scripts or tools for custom reporting and analysis.
5. What does "skill audit" mean in Skilled? The skill audit is an automated analysis that applies predefined heuristics to your skill usage history. It identifies trends like "Rising" skills (significant increase in use), "Stale" skills (not used recently), and "Heavy Hitters" (most frequently used), helping you manage and refine your AI tool repertoire.