📊 spec-kit-analytics

Measure what your AI builds, and how much time it saves you.

spec-kit-analytics is a Spec Kit extension that silently collects performance metrics in the background while you work. Every command you run is timed, and an LLM evaluates how long a human would have taken to produce the same result, giving you hard numbers on your productivity gains.

This is just the beginning. The analytics story is much bigger than timers and estimates: cost tracking, chat history logging, iteration counts, and a live dashboard are all on the roadmap. We're building the definitive measurement layer for AI-assisted software development, and we're shipping fast. Stay tuned. 🚀

---

✨ Features

Feature	Description
Execution Timer	Automatic start/end timestamps and readable durations for every command
Human Time Estimator	LLM-powered estimates of how long a junior & senior dev would take
Task Projection	After tasks, estimates implementation time per task
Format Toggle	Output in Markdown (default) or JSON
Gitignore Manager	One-click .gitignore protection for analytics data
Purge Command	Delete analytics for one spec or all at once
Show Command	Print collected stats. No file opening required

📖 Full documentation →

---

🚀 Installation

specify extension add analytics

or

specify extension add analytics --from https://github.com/Fyloss/spec-kit-analytics/archive/refs/tags/<VERSION>.zip

That's it. Analytics start collecting immediately with sensible defaults. Edit .specify/extensions/analytics/analytics-config.yml to customize.

---

⚙️ Configuration

Config file: .specify/extensions/analytics/analytics-config.yml

times:
  splitByCommands: false
humanEstimates:
  commands: true
  splitByCommands: false
  tasks: true
gitignore: false
# format: markdown

→ Full configuration reference, format toggle & gitignore manager

---

📁 Analytics Output Structure

<spec-folder>/
  analytics/
    time/
      commands.md              # (or .json, or commands/<name>.md if splitByCommands)
      humanEstimates/
        commands.md            # Per-command human time estimates
        tasks.md               # Task implementation time projections

---

🧹 Purge Command

analytics.purge <spec-name>  # Delete analytics for a specific spec
analytics.purge --all        # Delete all analytics across the project

→ Purge command docs

---

👀 Show Command

A pure local renderer. Print collected stats.

analytics.show <spec-name>   # Print stats for a specific spec
analytics.show --all         # Print stats for every spec, with project totals
analytics.show               # Same as --all

→ Show command docs

---

🗺️ TODO / Roadmap

• Interactive setup wizard: a guided configuration prompt at install time. Requires an on_install hook in Spec Kit itself (not yet available); we may propose this upstream.
• Cost calculations based on model usage (token counts × pricing).
• Chat history logging divided by commands.
• Iterations count tracking: number of messages/commands to the agent per session.
• Metrics visualization: a command to create a lightweight dashboard.

---

📄 License

MIT

---