Use when the user explicitly asks to regenerate, rebuild, or refresh the `claude-prospector` token-usage dashboard with no interpretive question attached — runs `python -m claude_prospector dashboard`, writes the HTML file, and reports the output path. Do NOT use when the user is asking what the data means, where their budget stands, or what to change (use `usage-analysis` instead — it regenerates as part of answering). Trigger phrases: "regenerate the usage dashboard", "rebuild my claude-pro...
Scanned 9/22/2026
Install to Claude Code
npx -y skills add glitchwerks/claude-prospector --skill usage-dashboard --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Usage Dashboard?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/glitchwerks-usage-dashboard)More formats (shields.io, HTML) on the badges page.
---
name: usage-dashboard
description: >
Use when the user explicitly asks to regenerate, rebuild, or refresh the
`claude-prospector` token-usage dashboard with no interpretive question
attached — runs `python -m claude_prospector dashboard`, writes the HTML
file, and reports the output path. Do NOT use when the user is asking what
the data means, where their budget stands, or what to change (use
`usage-analysis` instead — it regenerates as part of answering).
Trigger phrases: "regenerate the usage dashboard", "rebuild my
claude-prospector dashboard", "refresh the token dashboard", "regen the
prospector dashboard".
---
## Prerequisites
This skill invokes `python -m claude_prospector` under the hood. The Python package must be installed in the environment Claude Code uses. See the [README install steps](https://github.com/glitchwerks/claude-prospector#install-as-a-claude-code-plugin) for the two-step install.
# Usage Dashboard Skill
You are regenerating the `claude-prospector` interactive HTML token-usage dashboard for the user. The user triggered this skill explicitly — produce the dashboard and report the output path. **Do not interpret the data.** The conversational analysis path lives in the `usage-analysis` skill; this skill is the bare "regenerate the file" surface.
## Companion skill
For interpretation (budget status, top consumers, recommendations), point the user at `usage-analysis` instead. If the user's request mixes "rebuild the dashboard" with "and tell me what's happening", run this skill first to refresh the file, then hand off to `usage-analysis` for the analysis pass.
## Default behavior
If the user passes no arguments, regenerate against the **full aggregated session history** and write the dashboard to a known path. Pass `--no-open` so the file lands at the known path without spawning a browser tab — that side effect is usually unwanted from inside a Claude Code session.
When `$CLAUDE_PLUGIN_DATA` is set (the normal plugin context), the CLI automatically defaults `--output` to `$CLAUDE_PLUGIN_DATA/dashboard.html` and creates the parent directory if needed — you can omit `--output` entirely:
```bash
python -m claude_prospector dashboard --no-open
```
When running outside the plugin context (`$CLAUDE_PLUGIN_DATA` is unset), pass `--output` explicitly to control where the file lands:
```bash
# POSIX
python -m claude_prospector dashboard --output "$HOME/.claude/claude-prospector/dashboard.html" --no-open
# Windows
python -m claude_prospector dashboard --output "%USERPROFILE%\.claude\claude-prospector\dashboard.html" --no-open
```
The CLI prints `Dashboard written to <path>` to stdout on success. Echo that line back to the user so they have the absolute path to open or read.
## Arguments
The user may pass any of these. Forward them through to the underlying CLI; do not re-interpret.
| Argument | Behavior |
| --- | --- |
| `--window 5h` | 5-hour rolling window — scopes the dashboard to the last 5 hours. |
| `--window 7d` | 7-day rolling window — scopes the dashboard to the last 7 days. |
| `--window 30d` | 30-day rolling window — useful for monthly trend context. Larger windows are slower; flag that if `--window` exceeds 30d. |
| `--from YYYY-MM-DD --to YYYY-MM-DD` | Explicit date range. Mutually exclusive with `--window`. If both are supplied, surface the conflict and ask which one to use. |
| Any other CLI flag (`--limit-5h`, `--limit-7d`, `--limit-sonnet-7d`, `--format json`, `--data-dir`, etc.) | Forward verbatim. See `python -m claude_prospector dashboard --help` for the full surface. |
Always include `--no-open` in the final command line. When `$CLAUDE_PLUGIN_DATA` is set (the normal plugin context), omitting `--output` is fine — the CLI defaults to `$CLAUDE_PLUGIN_DATA/dashboard.html` automatically. Only pass `--output <path>` explicitly when the user requests a custom path or when running outside the plugin context.
## When the dashboard already exists
The CLI overwrites `--output` unconditionally. Do not pre-check or back-up the existing file — that's the user's job if they want one. Just regenerate.
## Errors
If the CLI exits non-zero, capture stderr and surface it to the user verbatim under a `**Dashboard generation failed:**` heading. Common causes:
- `python` not on PATH — the user has not installed the Python package; point them at the README's `## Install as a Claude Code plugin` § `### Prerequisite: Python package` section.
- `ModuleNotFoundError: claude_prospector` — same cause as above; the plugin is installed but the package isn't.
- Permission errors on the output path — surface verbatim and suggest passing a different `--output`.
Do not retry; surface the error and stop.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!