clawtel setup

Get clawtel reporting token usage from any project that calls Anthropic (SDK, Claude Code, openclaw, staffchief-style agents) to the claw.tech leaderboard.

clawtel is a single Go binary that reads aggregate token counts from a local tapes SQLite database and sends heartbeats to https://ingest.claw.tech/v1/heartbeat. It never reads prompt or response content — only created_at, model, prompt_tokens, completion_tokens from the nodes table.

When to use

• A project is calling the Anthropic API (directly via SDK, via Claude Code, or via any agent framework) and you want its usage on the claw.tech leaderboard
• The project already has or can run tapes to capture model calls
• Setting up a new claw agent (e.g. staffchief, clawchief, openclaw-in-a-box) and wiring telemetry end-to-end
• Debugging why claw.tech shows no heartbeats for a running agent

Prerequisites

Before running clawtel you need three things:

1. tapes installed and running as a proxy in front of Anthropic. If tapes isn't there yet:

   curl -fsSL https://download.tapes.dev/install | bash
   tapes init
   

2. A claw.tech ingest key. Register your claw at https://claw.tech and copy the ik_... key shown once at creation.
3. An Anthropic-calling workload. Could be a Node/Python app importing @anthropic-ai/sdk / anthropic, Claude Code, or an openclaw agent. Anything that hits api.anthropic.com.

Step 1 — wire tapes in front of Anthropic

clawtel only sees usage that tapes has recorded. Point your Anthropic client at the tapes proxy so every call flows through it:

# In the shell where your agent runs
export ANTHROPIC_BASE_URL="http://localhost:8080"
tapes start          # starts proxy + API, writes to ~/.tapes/tapes.sqlite

For the Anthropic Node/Python SDK, ANTHROPIC_BASE_URL (or passing baseURL to the client constructor) is enough — no code changes. For Claude Code, set it before launching the CLI.

OpenClaw users: ANTHROPIC_BASE_URL alone is not enough

If your workload is an OpenClaw agent (clawchief, staffchief, openclaw-in-a-box), the env var is silently ignored. OpenClaw instantiates its Anthropic client with baseURL: model.baseUrl, which clobbers the SDK's normal readEnv("ANTHROPIC_BASE_URL") fallback. You'll see heartbeats sending with model="", input_tokens=0, output_tokens=0 forever, and the gateway process will hold a direct TLS connection to Anthropic's edge instead of 127.0.0.1:8080.

Fix: set the provider's base URL explicitly in ~/.openclaw/openclaw.json:

{
  "models": {
    "providers": {
      "anthropic": {
        "baseUrl": "http://localhost:8080",
        "models": [
          { "id": "claude-opus-4-6",   "name": "Claude Opus 4.6"   },
          { "id": "claude-sonnet-4-6", "name": "Claude Sonnet 4.6" },
          { "id": "claude-haiku-4-5",  "name": "Claude Haiku 4.5"  }
        ]
      }
    }
  }
}