install-skill-tracker_skill

This skill installs a global Claude Code tracking system to log skill usage, duration, token costs, and prompts for insight.
  • Python

14

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill dmccreary/claude-skills --skill install-skill-tracker

  • README.md23.9 KB
  • SKILL.md11.2 KB

Overview

This skill installs a global Claude Code skill tracking system that logs every skill invocation across all projects. It captures execution start/end, duration, token usage, and the user prompt that triggered each skill for centralized analysis. Use it to surface costly or slow workflows and to discover opportunities for new automated skills.

How this skill works

The installer creates a ~/.claude/ directory structure and places hook scripts that run on UserPromptSubmit, PreToolUse, and PostToolUse events. Hooks write compact JSONL entries to ~/.claude/activity-logs/ (prompts.jsonl and skill-usage.jsonl) containing timestamps, session IDs, project paths, token metrics, and durations. Included analysis scripts parse those logs to produce frequency, timing, token-cost, and cache-efficiency reports.

When to use it

  • Set up centralized tracking across all Claude Code projects.
  • Measure skill execution time to find slow or expensive steps.
  • Monitor token usage and estimate API costs per skill.
  • Identify repeated prompts or workflows for new skill creation.
  • Audit prompt cache effectiveness and optimization opportunities.

Best practices

  • Install globally in ~/.claude/ so logs aggregate across projects.
  • Merge hook entries into an existing ~/.claude/settings.json instead of overwriting.
  • Make hook and analysis scripts executable (chmod +x).
  • Verify timing by running a sample skill and checking skill-usage.jsonl.
  • Keep logs local and rotate or delete activity-logs periodically for privacy or disk management.

Example use cases

  • Find which skills consume the most tokens and estimate monthly API spend.
  • Discover long-running skills to prioritize for optimization or parallelization.
  • Surface repeated manual prompts that can be turned into new, reusable skills.
  • Compare session activity timelines to identify distraction gaps or workflow bottlenecks.
  • Validate prompt cache behavior by measuring cache read vs creation tokens.

FAQ

No. All logs are stored locally in ~/.claude/activity-logs/ and are not transmitted externally by default.

What if my existing settings.json already configures hooks?

Merge the new hook definitions into your existing ~/.claude/settings.json rather than replacing the file to preserve other configurations.

Why do I see JSON parsing errors?

Ensure hooks emit compact JSONL using jq -nc and fix existing pretty-printed files with jq -c '.' to restore valid JSONL format.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
install-skill-tracker skill by dmccreary/claude-skills | VeilStrat