skill-telemetry
Client-side telemetry reporter. When any Skill is activated, this Skill sends a single invocation event to skill-telemetry-server for usage analytics.
Privacy: Only Skill metadata (name, version, author) and machine identifiers (OS, IP) are reported. No user input, file content, or secrets are collected. All data stays on the internal network.
When to Use
Triggered automatically by the ## Telemetry (MUST) section in every Skill. No user interaction required.
Input
| Param | Required | Source | Description |
|---|---|---|---|
name |
✅ | Caller's frontmatter name |
Skill identifier |
version |
✅ | Caller's metadata.version |
Skill version |
author |
✅ | Caller's metadata.author |
Skill author |
Agent Execution Rule
Agent MUST run the telemetry command below BEFORE executing the caller Skill's logic. Mandatory — must not be skipped. Failure is non-blocking but the attempt is required.
Replace {{SKILL_NAME}}, {{SKILL_VERSION}}, {{SKILL_AUTHOR}} with actual values from the caller Skill's frontmatter.
The agent MUST resolve the path to
scripts/telemetry.py(orscripts/telemetry.ps1on Windows) relative to this SKILL.md file. If the path cannot be resolved, setSKILL_TELEMETRY_SCRIPTto the absolute path. If both fail, skip telemetry silently.
Linux / macOS:
python3 "${SKILL_TELEMETRY_SCRIPT:-<resolve scripts/telemetry.py>}" "{{SKILL_NAME}}" "{{SKILL_VERSION}}" "{{SKILL_AUTHOR}}" 2>/dev/null &
Windows:
& "${env:SKILL_TELEMETRY_SCRIPT:-<resolve scripts/telemetry.ps1>}" '{{SKILL_NAME}}' '{{SKILL_VERSION}}' '{{SKILL_AUTHOR}}'