Observability and cost tracking for Cline
See every Cline task, message, tool call, subagent and token as it happens. ClawMetry finds Cline on your machine on its own and renders tasks, transcripts, cache-aware token splits and cost per task in one dashboard, next to every other agent you run.
What ClawMetry sees on Cline
Cline is the open source coding agent from Cline Bot Inc., Apache-2.0 licensed and built in the
open by hundreds of contributors, with roughly 66.6k stars on
GitHub. You run it as a
VS Code extension, as a JetBrains plugin, inside Cursor, Windsurf, Antigravity or Zed, through
Neovim in ACP mode, or headless in the terminal with npm i -g cline. It is model
agnostic, so one week of tasks can span Anthropic, OpenAI, Gemini, Bedrock, Vertex, OpenRouter and
a local Ollama model. Once ClawMetry is running, that activity shows up as:
- Tasks per project: one row per Cline task, labelled with the directory it ran in, the model and provider behind it, when it started, when it last moved, its message count and its token totals.
- The full transcript: your instruction, Cline's replies, its reasoning where it recorded any, and every tool call with its arguments and its result, in the order they happened. File edits, terminal commands, browser steps and the tools your own MCP servers expose all read the same way.
- Subagent fan-out, priced: Cline decides on its own when to launch parallel research subagents, each with its own prompt and its own token budget. ClawMetry nests each one under the turn that launched it, showing the brief it was handed, the report it sent back, and its own tokens and cost. A research burst reads as several priced children, not one unexplained jump in the parent.
- Token splits that include the cache: fresh input, cached input, cache writes and output, counted per step and added up per task, so a long autonomous run is not reported as a single call.
- Cost priced per provider: because Cline lets you swap models mid-project, the dollar figure follows the provider and model that actually ran each task. A task on a local model is shown at zero rather than priced as though someone had billed you for it.
- Approvals in context: what Cline asked permission for, what you answered, and how long the task sat waiting for you before it moved.
- Rules and skills: the rules and skills Cline loads, global and per project, in the same browser as every other runtime's.
- One runtime, not three: the extension, the JetBrains plugin and the CLI share the same agent core, so ClawMetry keeps them in one Cline view instead of splitting your history by which surface you happened to open.
Enable it in about two minutes
1. Install ClawMetry
Pick the line that matches your OS. Windows users: paste the install.cmd line, not the shell one.
# macOS or Linux
curl -fsSL https://clawmetry.com/install.sh | bash# Windows (cmd.exe or PowerShell)
curl -fsSL https://clawmetry.com/install.cmd -o install.cmd && install.cmdOr skip the terminal entirely and grab the desktop app: macOS (.dmg) · Windows (.exe) · Linux (.AppImage). It installs and updates ClawMetry for you.
2. Run the daemon
clawmetryThe daemon finds Cline on the machine and starts reading immediately. No config, no editor restart, and nothing to add to your Cline setup.
3. Open the dashboard
Open http://localhost:8900. Switch the runtime chip to Cline.
Start the free trial →The one thing this catches that Cline alone does not
Cline is honest about the task in front of you: the sidebar shows that task's tokens and what it cost. What no single task view can show you is the shape of the week. How many tasks you left running across how many repos and how many editors. Which ones burned their budget re-reading the same files instead of editing them. How much of yesterday's spend went to research subagents whose findings you never used. Whether the model you quietly switched to last Tuesday is cheaper or just slower. And how Cline's total compares with the Claude Code and Codex runs happening beside it on the same laptop. ClawMetry keeps that view: cost per hour across every task and every runtime, subagents billed separately instead of hidden inside a parent, repeated-call detection, and token-growth curves per task. Set a threshold and the approval queue pauses before the next call. Kill switch (Pro) stops it.
Starred by the people building AI
Engineers and founders from OpenAI, Google, PostHog and more have starred ClawMetry on GitHub. One of them helped build it.
All 26 runtimes ClawMetry observes
Cline is one of 26. Free on every plan: OpenClaw, NVIDIA NemoClaw. Paid on Cloud Pro or Self-Hosted Pro: the rest.