Spike: Investigate usage tracking for ai-helpers skills and agents

[patternfly-jira-sync]

• Investigate what usage data can be collected for ai-helpers marketplace skills and agents across Claude Code and Cursor

• Evaluate feasibility of each tracking layer: marketplace install counts, Claude Code OpenTelemetry (built-in), and any cross-platform approaches

• Security is the top priority — no PII, no risk of data exposure, developer trust must be preserved

• Determine what metrics would help answer questions like: How many teams have adopted? Which skills are used most? Is usage growing over time?

• Assess whether the existing OpenShift → Segment → Amplitude pattern (used for Forklift) can be adapted, or if Claude Code's built-in OTel is sufficient

• Research external standards and Cursor's telemetry capabilities

• Where applicable, include recommendations for how related tooling (e.g. MCP servers) could complement the tracking strategy

• Deliverable: recommendation document with feasibility assessment, proposed architecture, privacy constraints, and recommended metrics

Acceptance Criteria:

• Document which tracking layers are feasible, including cross-platform coverage (Claude Code vs Cursor)

• Identify security and privacy constraints — no-go any approach that requires PII or risks trust

• Propose metrics that answer: how many teams adopted, which skills are most used, and is usage growing

• Recommend an architecture (OTel → collector → Amplitude, or simpler)

• Include comparison to how GitHub Copilot and VS Code extensions handle usage tracking

---

Jira Issue: PF-4379