How does the skill CLI implement telemetry?

How does the skill cli implement telemetry?

Excellent! Now I have a comprehensive understanding. Let me create a detailed summary:

##How the Skills CLI Implements Telemetry

The Skills CLI implements telemetry using a simple, non-intrusive approach with fire-and-forget HTTP requests. Here's how it works:

###Core Implementation (src/telemetry.ts)

The telemetry system is built on these key components:

1. Telemetry URL: All data is sent to https://add-skill.vercel.sh/t

2. Event Types: Four main events are tracked:

   • install - When skills are installed (tracks source, skills installed, target agents, global install flag, etc.)
   • find - When users search for skills (tracks query, result count, interactive mode)
   • check - When users check for skill updates (tracks skill count and updates available)
   • update - When skills are updated (tracks total skills, success count, fail count)

3. Enabling/Disabling: Telemetry respects user preferences:

   • Disabled via DISABLE_TELEMETRY environment variable
   • Disabled via DO_NOT_TRACK environment variable
   • Automatically detects CI environments (GitHub Actions, GitLab CI, CircleCI, Travis, BuildKite, Jenkins, TeamCity) and includes a ci: '1' flag

4. Data Parameters: Every request includes:

   • v - CLI version (from package.json)
   • ci - Set to '1' if running in a CI environment
   • Event-specific data as URL parameters

###Non-Intrusive Design

The implementation emphasizes user safety and trust:

• Fire-and-forget: Telemetry is sent asynchronously with fetch() and no error handling—it will never break the CLI
• Silent failures: All errors are caught and silently ignored (catch(() => {}))
• No blocking: The CLI continues immediately without waiting for telemetry responses
• Version tracking: Set via setVersion() during CLI initialization

###Telemetry Usage in Commands

Install Command (src/add.ts):

track({
  event: 'install',
  source: remoteSkill.sourceIdentifier,  // e.g., "mintlify/bun.com"
  skills: remoteSkill.installName,       // comma-separated skill names
  agents: targetAgents.join(','),        // target agents
  global: installGlobally ? '1' : undefined,
  skillFiles: JSON.stringify({ [name]: url }),
  sourceType: remoteSkill.providerId,    // 'mintlify', 'github', 'huggingface', etc.
});

Find Command (src/find.ts):

• Non-interactive: { event: 'find', query, resultCount }
• Interactive: { event: 'find', query: '', resultCount, interactive: '1' }

Check Command (src/cli.ts):

track({
  event: 'check',
  skillCount: String(checkRequest.skills.length),
  updatesAvailable: String(data.updates.length),
});

Update Command (src/cli.ts):

track({
  event: 'update',
  skillCount: String(updates.length),
  successCount: String(successCount),
  failCount: String(failCount),
});

###Initialization

The CLI version is loaded from package.json at startup in src/cli.ts:

const VERSION = getVersion();  // Reads from package.json
initTelemetry(VERSION);        // Sets version for all subsequent telemetry calls

This design ensures minimal performance impact while providing Vercel Labs with usage metrics to understand how skills are being discovered, installed, and updated.