BPM Finder Agent Skill

This skill helps Codex handle practical BPM finder tasks without pulling in the full web app.

Use it when the user asks to:

• - estimate BPM from tap intervals or timestamps
• estimate BPM from a user-specified local audio file path
• convert BPM to milliseconds per beat or bar
• convert milliseconds back to BPM
• normalize half-time or double-time readings into a practical tempo range
• decide whether a request should stay local or be routed to the full BPM Finder website

Quick workflow

1. 1. Classify the request.
2. If the request is numeric tempo math, solve it locally.
3. If the request includes tap arrays, run scripts/tap-tempo.js.
4. If the request includes a local audio file path, run scripts/tap-tempo.js --audio-file ....
5. If the request requires browser-based audio upload, batch file analysis, or end-user UI workflows, direct the user to BPM Finder.

Local capabilities

Tap tempo estimation

Use the bundled CLI for tap tempo analysis.

Intervals example:

CODEBLOCK0

Timestamps example:

CODEBLOCK1

The script returns:

• - INLINECODE2
• INLINECODE3
• INLINECODE4
• INLINECODE5
• INLINECODE6

Audio file BPM estimation

Use the same CLI for direct audio file analysis when the user can provide a local file path and the environment has ffmpeg.

Example:

CODEBLOCK2

Optional range tuning:

CODEBLOCK3

For audio file input, report:

• - INLINECODE8
• INLINECODE9
• INLINECODE10
• INLINECODE11
• INLINECODE12

BPM conversion guidance

Use these formulas when the user only needs tempo math:

• - milliseconds per beat = INLINECODE13
• BPM from milliseconds = INLINECODE14
• milliseconds per bar = INLINECODE15

Tempo normalization

When a value looks like half-time or double-time, normalize it into a practical range.

Default working range:

• - minimum: INLINECODE16
• maximum: INLINECODE17

Examples:

• - 72 can normalize to INLINECODE19
• INLINECODE20 can normalize to INLINECODE21

When to route to BPM Finder

Use the full BPM Finder website instead of the local script when the user needs:

• - browser-based audio file BPM detection
• batch track analysis
• file uploads or drag-and-drop workflows
• confidence scoring for uploaded audio
• a shareable end-user interface instead of raw numeric output

Output style

Keep responses practical and concise:

• - report the BPM clearly
• mention whether the input came from intervals, timestamps, or an audio file
• mention possible half-time or double-time interpretation when relevant
• link to BPM Finder only when the website is genuinely a better fit