claude-performance-tracker

v0.7.1

Qualify good agent usage and compare approaches on a cost-per-successful-outcome basis (tokens, time, prompts).

Claude Code8 Skills

By Diego Zuluaga0 GitHub starsUpdated 4 days ago

Directory evidence

Runtimes: Claude Code
Parsed components: 8 skill or MCP entries
Source updated: Sep 19, 2026
Manifest status: Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology →

Install claude-performance-tracker for Claude Code

Installs for the current user

claude plugin marketplace add IchenDEV/agent-plugin-mkt
claude plugin marketplace update agent-plugin-marketplace
claude plugin install claude-performance-tracker@agent-plugin-marketplace

Paste and run these commands in a terminal with Claude Code. They add and refresh the PluginsMP catalog, then install this plugin.

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually

$ git clone https://github.com/Muroz/agent-toolbox

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The plugin root is plugins/claude-performance-tracker/.

Plugin files

plugins/claude-performance-tracker/

├── .claude-plugin/plugin.json
├── skills/evaluate-run/SKILL.md
├── skills/track/SKILL.md
├── skills/track-done/SKILL.md
├── skills/track-list/SKILL.md
├── skills/track-pause/SKILL.md
├── skills/track-resume/SKILL.md
├── skills/usage-lessons/SKILL.md
└── skills/usage-report/SKILL.md

Included Skills8

evaluate-runskills/evaluate-run/SKILL.md

Score the qualitative side of one or more runs using the usage-evaluator subagent — agent-behavior rubric plus per-prompt quality. Use to evaluate recent runs or a specific run id. Runs out of the hot path (opt-in / batched).

trackskills/track/SKILL.md

Open a deliberately-tracked performance run for the work you are about to do. Use when you want to measure and later compare an approach (model, plan-mode, subagent, skill) against others on a real task. Asks for a task label, type, size and intended approach.

track-doneskills/track-done/SKILL.md

Close the currently-open tracked performance run and record its outcome. Use when the task you bracketed with /track is finished (or abandoned). Asks for outcome and a 1-5 satisfaction score.

track-listskills/track-list/SKILL.md

List the open tracked performance runs and their state — which is active (and in which session) and which are paused. Use to see what can be resumed or closed when juggling multiple tracked tasks across sessions.

track-pauseskills/track-pause/SKILL.md

Pause the tracked performance run this session is driving, without finalizing it. Use when you need to set the current task aside to work on something else, then resume it later (here or in another session). Keeps the run open and resumable.

track-resumeskills/track-resume/SKILL.md

Resume a previously paused tracked performance run and attach it to this session. Use to pick a task back up — including in a different session from where it started — so its remaining work is measured on the same run. Identifies the run by id or task label.

usage-lessonsskills/usage-lessons/SKILL.md

Synthesize a durable, git-shareable usage playbook from all your tracked runs — what approaches work, recurring friction, weak prompt habits and drift. Uses the lessons-synthesizer subagent. Runs out of the hot path.

usage-reportskills/usage-report/SKILL.md

Report on your tracked Claude usage — overview, approach comparison, model-degradation trend or a single run's scorecard. Use to understand cost (tokens/time/prompts) per successful outcome and to compare how different approaches fare.

Plugin manifests1

plugins/claude-performance-tracker/.claude-plugin/plugin.json

{
  "name": "claude-performance-tracker",
  "version": "0.7.1",
  "description": "Qualify good agent usage and compare approaches on a cost-per-successful-outcome basis (tokens, time, prompts).",
  "author": {
    "name": "Diego Zuluaga"
  },
  "keywords": [
    "usage",
    "metrics",
    "telemetry",
    "evaluation",
    "observability"
  ]
}