Private AI coding telemetryClaude Code autosync todayCodex manual sync supportedSDK + MCP for custom workflows
See how your team is actually using AI coding tools.
burnlog gives teams a shared view of token usage, source mix, models, sessions, and estimated spend across Claude Code, Codex log imports, and custom agent workflows — without syncing prompt content off-machine.
Tracked developers
8
Already syncing usage telemetry
Tracked tokens
2.6B
Across shared burnlog boards
Collection scope
Usage metadata
Token counts and session stats — not prompt bodies
burnlog is best used as workflow telemetry and monitoring. Public leaderboard and challenge numbers are self-reported unless your team wires burnlog into a trusted runtime, so treat them as useful signals — not standalone proof.
WHAT SUPPORTS TODAY
A truthful path for each setup
Claude Code
Autosync with the hook + daemon
Native sync path today for teams that want usage to land automatically.
Codex
Manual scan and sync
Supported today through log scans and manual sync, with the daemon as an optional local helper.
SDK + MCP
Custom workflows and readback
Instrument your own agents or expose burnlog context inside supported runtimes.
WHAT TEAMS SEE
Useful signals, not vanity charts
✦Token usage, session volume, and estimated spend over time
✦Source and model mix across supported tools and providers
✦A shared board the team can review without exposing prompt content
WHY IT MATTERS
Visibility across real workflows
Put Claude Code autosync, Codex imports, and custom agent telemetry on one surface so leads can compare adoption without chasing screenshots or local logs.
WHY IT MATTERS
Privacy-first by design
burnlog syncs usage metadata, token counts, and related session stats — not prompt bodies — so teams get narrower collection by default.
WHY IT MATTERS
Useful for ops and coaching
Track model concentration, weekly momentum, and estimated spend to understand how AI usage changes as your team adopts new tools and habits.
WORKS TODAY
Pick the setup that matches your stack
Start with native Claude Code autosync, use the current manual Codex flow, or wire burnlog into your own agent runtime with the SDK or MCP server.
Claude Code
Install hook · autosync
$ burnlog install && burnlog daemon
Best path today. Install the Claude Code hook, keep the local daemon running, and burnlog syncs usage automatically as sessions land.
Codex
Scan · sync · daemon
$ burnlog scan && burnlog sync
Codex is supported through a manual log scan/sync flow today. Use the daemon if you want a local process watching for new logs between syncs.
Custom agents
SDK instrumentation
$ import { Burnlog } from '@sxna/burnlog-sdk'
Use the SDK to emit token usage from your own agent loops, background jobs, or internal tooling without changing the dashboard your team already sees.
In-agent readback
MCP access
$ Connect the burnlog MCP server
Expose prior usage and profile stats inside supported agent runtimes so the agent can read back burnlog context without custom glue code.
8 developers currently syncing 2.6B tracked tokens into burnlog
◆Your prompts stay local. burnlog syncs usage metadata and token counts, never prompt content. Open source CLI — audit it yourself.Privacy →
Total Burned
2.6B
across all users
This Week
0
last 7 days
Active Burners
8
competing now
Avg / User
321.3M
all-time average
Leaderboard
#UserTrendTokensRank
