vs ccusage / CodeBurn

How wigtoken compares

Three popular options for tracking Claude Code usage, and where each lands.

wigtokenccusageCodeBurn
SurfaceServer + dashboard + npm widgetCLIHosted web app
HostingSelf-host (any)Local onlySaaS (their infra)
Multi-user❌ (single user)
Multi-machine✅ (agent or hook)
Real-time✅ SSEn/a (one-shot)poll-based
Embed on public site✅ widget npm package
Prometheus metrics/metrics
Grafana dashboard✅ provided
Custom widget components16+n/a
Cost breakdown✅ per user / model / machine✅ per model✅ per model
Audit log
Token scopesingest / read / embed / adminn/an/a
Data ownershipyours (SQLite file)yours (local)theirs
Setup time2min solo / 10min team30s30s
LicenseMITMITproprietary

When to use which

  • ccusage — you’re a solo dev, only want a CLI summary, don’t care about embedding or history beyond what your local Claude Code retains.
  • CodeBurn — you want a pretty hosted UI, are okay with their service holding your data, solo only.
  • wigtoken — you want a team or org view, want to embed live counters on your marketing site, want Prometheus integration, or just don’t want a third party holding your usage data.

Migration paths

From ccusage

ccusage doesn’t store data beyond what Claude Code itself keeps. Just point wigtoken at your ~/.claude/projects/ directory — it’ll backfill from existing JSONL files on first scan.

From CodeBurn

CodeBurn’s data is held by them; export isn’t available at time of writing. Run both side-by-side during transition, then switch when wigtoken has the history you need (usually 2–4 weeks).

Feature ideas we won’t add

  • Anthropic-API-spend tracking. Anthropic Console already covers this; we focus on Claude Code session spend.
  • Per-prompt content storage. Privacy red line — we only store usage counts and metadata.
  • Slack / Discord notifications out of the box. Wire them up via Prometheus Alertmanager — that’s the right abstraction.