Agent Builder 지원 범위 내 머신 셀프 호스팅 요금제 Cloud 로그인 디바이스 문서 블로그 쇼케이스 로드맵 엔터프라이즈

내 머신에서 실행 · 지시할 때만 작동

AI 에이전트를 위한
컨트롤 플레인

  • 모든 에이전트를 봅니다.
  • 모든 동작을 파악합니다.
  • 일어나선 안 될 일을 막습니다.

ClawMetry는 팀이 이미 사용 중인 26개 에이전트 런타임을, Claude Code와 Cursor부터 Codex와 OpenClaw까지, 그들이 이미 남기는 파일에서 직접 읽어냅니다. 에이전트에 추가할 것은 없습니다.

$ pip install clawmetry && clawmetry

7일간 모든 런타임, 카드 등록 없이. 이후 노드당 월 $9 또는 $19.

또는 데스크톱 앱 받기 macOS Windows Linux

Verbatim record, not a mockup. A real detection from the laptop ClawMetry was built on. The steps, the counts and the message are what the daemon wrote to disk.

154개월간 감지: 진행 없음 6건, 도구 반복 실패 6건, 루프 정체 3건
7 of 68Claude Code 세션에서 최소 1건이 감지됨. 과하지도, 드물지도 않습니다.
0설정 불필요. 설치 즉시 감지가 시작되며 발견된 모든 런타임을 포함합니다.

Supported agent runtimes

1회 설치, 4개월간 실측

비용이 많이 드는 에이전트가 아니라,
멈춰버린 에이전트를 표시합니다.

길고 비싼 세션이 오히려 최고의 결과물인 경우가 많아, 큰 숫자만 쫓는 도구는 엉뚱한 것을 잡아냅니다. ClawMetry는 에이전트가 실제로 무언가를 만들어내고 있는지를 봅니다. 같은 노트북, 4개월, 두 사례를 나란히 놓았습니다.

정상 작동 · 표시된 적 없음

$171.58

1,700 turns · 527,741 tokens · 0 detections

The most expensive session of the four months. It ran long and cost real money, and it was writing code the whole time. ClawMetry left it alone, which is the correct answer.

표시됨 · 정체

$175.10

7 sessions · 4% of all sessions · 21% of all spend

Every one caught for the same reason: tool calls with no file changes, the same failure repeating, or a loop that stopped advancing. Most of them cost less than the session on the left.

감지기는 경고할 뿐, 개입하지 않습니다. 해당 7개 세션은 모두 끝까지 실행됐습니다. 표시는 알림이지 트리거가 아닙니다. 사람이 직접 중단하거나, 직접 설정한 지출 한도를 에이전트가 넘지 않는 한 아무것도 멈추지 않습니다.

165 sessions and $826.47 of agent spend on one developer’s laptop, April to August 2026. One machine, not a customer aggregate. Yours will look different, which is the reason to read your own.

직접 확인해 보세요

하나의 화면. 모든 에이전트에서 일어나는 모든 일.

clawmetry - flow
ClawMetry dashboard

빈칸까지 표에 그대로

런타임별로 가능한 것들.

이 분야의 모든 업체가 가드레일을 말합니다. 여기 빈 칸까지 포함한 전체 매트릭스가 있습니다.

런타임 관찰 실행 전 차단 실행 중 종료 달러 기준 비용
OpenClaw, NVIDIA NemoClaw, Goosefree forever via proxy
Claude Codethe only pre-execution tool gate
Codex, OpenCode, Aider via proxy
Cursorone IDE process holds every session
GitHub Copilotbilled in vendor credits, not USD
NanoClaw, PicoClaw, Kimi CLIno model id on disk
Qwen, Hermes, Pi, Deep Agents, n8n, Antigravity,
Grok, QM, DeepSeek Harness, Exo
via proxy

실행 전 차단

Claude Code gets a real pre-execution gate: a PreToolUse hook parks the call before the tool fires. Everything routed through the ClawMetry proxy gets spend refused with a 429 before tokens are spent.

실행 중 종료

A gateway kill for OpenClaw, and SIGSTOP then SIGKILL against the process tree for the CLI runtimes, with a guard so a recycled pid is never the one we hit. macOS and Linux. Windows has no SIGSTOP, so pause and resume do not exist there.

사후 감사

Every event carries a SHA-256 chain over its immutable fields, per node, with a verifier command. Approval decisions land in an append-only log. Works on every runtime.

노트북에 데몬을 올리기 전에

머신에 미치는 영향과,
밖으로 나가는 데이터.

추정치가 아니라 같은 노트북에서 실측한 값입니다.

리소스 사용량

Local store, 165 sessions and 19,977 events145 MB
Per session, average~880 KB
Network destinations, cloud offnone
Inbound ports opened0
Localhost query server127.0.0.1 + token
Removalclawmetry uninstall

The query server binds to loopback only, with a one-time 32-hex token and a discovery file at mode 0600. It is why the dashboard can read the store without holding the writer lock.

정확히 무엇이 나가는가

transcriptspromptstool callsfile paths logsmemoryyour code

Never. These stay in the local file.

WITH CLOUD ON ↓
AES-256-GCM ciphertextnode idversioncounters

The key is generated on the node and never leaves it. Decryption happens in your browser. Compelled by a subpoena, we hand over a blob we cannot open, plus the node id, the version and some counters.

로컬 파일은 평문입니다

Everything is sealed in transit and at rest in our cloud, and it is not encrypted on your own disk. The DuckDB file carries default user permissions. Anything running as your user can read it.

마스킹은 실패 시 그대로 통과합니다

Secrets are fingerprinted at ingest by default, before disk. It is pattern-based, and on an internal error it keeps the original value rather than dropping data. It is tuned for credential shapes, not customer PII.

연결된 노드는 원격으로 중단될 수 있습니다

If you turn cloud on, Stop, Pause and Resume are reachable from the dashboard and the desk device for any node on the account. That is the feature. Run local-only and no such channel exists.

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.

사용자 후기

ClawMetry에 대한 커뮤니티의 이야기

전체 보기
Simen.ai
Blog review
Blog
ClawMetry has become the go-to open source observability dashboard for OpenClaw agents. No configuration required, and the live flow visualization shows you the actual decision path your agent took, step by step.
Mykola Kondratiuk
Product Hunt
Product Hunt
Oh this is exactly what I needed. The biggest pain point has been figuring out where my tokens are going. Especially with sub-agents spawning other sub-agents, costs spiral fast and you have zero visibility.
@oadiaz
Medium
Medium
It's like going from 'I think my agents are working' to having a mission control center. The cost tracking alone was worth it. I discovered one agent was using 3x more tokens than necessary.
Product Hunt
#5 Launch Day
ClawMetry launched alongside Claude Sonnet 4.6 and held #5 all day. Featured in the PH newsletter under 'Keep agents in line.' 198+ upvotes.
Mayank Jain
LinkedIn
LinkedIn
What are your AI agents actually doing behind the scenes? Most builders don't know. They just hope everything works. But hope is not observability. Meet ClawMetry.
awesome-openclaw
Community list
Awesome List
Open source observability for OpenClaw: token costs, session drift, memory alerts. Nothing leaves your machine.
FunBlocks AI Reviews
Review
Review
A must-have for any serious developer or startup working with AI sandboxes.
@jonah_lipsitt
▶ Video
I got tired of not knowing what my OpenClaw was doing. A Phillips Hue spotlight now shines on my Mac Mini based on what it is doing, further visualized by @clawmetry
@KalariyaHiren1
Thread
Ask OpenClaw to do something. It starts working. And you just... wait. Hoping. Is it searching? Which websites? Where is it stuck right now? Zero visibility. ClawMetry fixes that.
@Save_Delete
If you are building with OpenClaw, ClawMetry is the missing piece. Real-time visibility into every sub-agent, tool call, token cost, and session log.
Mihail / HookWatch
Product Hunt
Product Hunt
This is exactly the kind of tooling the AI agent ecosystem needs right now. The per-session cost breakdown is a smart move. Most teams have no idea what their agents actually cost until the invoice hits.
@nickvnturi
This looks like a very helpful tool for tracking agents. Thanks for sharing it.
@Asif2BD
ClawMetry and our JARVIS Mission Control complement each other. Observability and coordination are separate problems.
@complete_ai_training
Instagram
Instagram
On the 19th of February, I discovered 5 exciting new AI tools you won't want to miss: ClawMetry for OpenClaw, Flixier, Your AI Clone, Baseline Core, and STUD.
@ProductHunToday
Top 5 · Product Hunt
Top 5 on Product Hunt (Feb 18, 2026): #5 ClawMetry for OpenClaw, launched alongside Claude Sonnet 4.6.
Harsh Upadhyay
Product Hunt
Product Hunt
Really like the clarity of 'Know what your agents are doing. Right now.' The core pain you're solving feels bigger than a single ecosystem. Very clean execution.
Damian Saez
Product Hunt
Product Hunt
Observability for AI agents is something most builders overlook until things break in production. Love that you're tackling this early.
Simen.ai
Blog review
Blog
ClawMetry has become the go-to open source observability dashboard for OpenClaw agents. No configuration required, and the live flow visualization shows you the actual decision path your agent took, step by step.
Mykola Kondratiuk
Product Hunt
Product Hunt
Oh this is exactly what I needed. The biggest pain point has been figuring out where my tokens are going. Especially with sub-agents spawning other sub-agents, costs spiral fast and you have zero visibility.
@oadiaz
Medium
Medium
It's like going from 'I think my agents are working' to having a mission control center. The cost tracking alone was worth it. I discovered one agent was using 3x more tokens than necessary.
Product Hunt
#5 Launch Day
ClawMetry launched alongside Claude Sonnet 4.6 and held #5 all day. Featured in the PH newsletter under 'Keep agents in line.' 198+ upvotes.
Mayank Jain
LinkedIn
LinkedIn
What are your AI agents actually doing behind the scenes? Most builders don't know. They just hope everything works. But hope is not observability. Meet ClawMetry.
awesome-openclaw
Community list
Awesome List
Open source observability for OpenClaw: token costs, session drift, memory alerts. Nothing leaves your machine.
FunBlocks AI Reviews
Review
Review
A must-have for any serious developer or startup working with AI sandboxes.
@jonah_lipsitt
▶ Video
I got tired of not knowing what my OpenClaw was doing. A Phillips Hue spotlight now shines on my Mac Mini based on what it is doing, further visualized by @clawmetry
@KalariyaHiren1
Thread
Ask OpenClaw to do something. It starts working. And you just... wait. Hoping. Is it searching? Which websites? Where is it stuck right now? Zero visibility. ClawMetry fixes that.
@Save_Delete
If you are building with OpenClaw, ClawMetry is the missing piece. Real-time visibility into every sub-agent, tool call, token cost, and session log.
Mihail / HookWatch
Product Hunt
Product Hunt
This is exactly the kind of tooling the AI agent ecosystem needs right now. The per-session cost breakdown is a smart move. Most teams have no idea what their agents actually cost until the invoice hits.
@nickvnturi
This looks like a very helpful tool for tracking agents. Thanks for sharing it.
@Asif2BD
ClawMetry and our JARVIS Mission Control complement each other. Observability and coordination are separate problems.
@complete_ai_training
Instagram
Instagram
On the 19th of February, I discovered 5 exciting new AI tools you won't want to miss: ClawMetry for OpenClaw, Flixier, Your AI Clone, Baseline Core, and STUD.
@ProductHunToday
Top 5 · Product Hunt
Top 5 on Product Hunt (Feb 18, 2026): #5 ClawMetry for OpenClaw, launched alongside Claude Sonnet 4.6.
Harsh Upadhyay
Product Hunt
Product Hunt
Really like the clarity of 'Know what your agents are doing. Right now.' The core pain you're solving feels bigger than a single ecosystem. Very clean execution.
Damian Saez
Product Hunt
Product Hunt
Observability for AI agents is something most builders overlook until things break in production. Love that you're tackling this early.
Product Hunt에서 모든 댓글 보기

실행 위치는 세 가지

Cloud는 선택 사항입니다.
처음부터 그랬습니다.

데몬은 사용자 디스크의 DuckDB 파일로 계산 결과를 씁니다. 그 이후 대시보드를 어디에 둘지는 선택이며, 어느 쪽을 골라도 요금은 같습니다.

01 · Local only

아무것도 나가지 않음

Included in the MIT core

The dashboard reads the local file through a loopback query server. There is no account, no sync and no destination to point at.

  • Network destinations: none
  • Inbound ports opened: 0
  • Free forever for OpenClaw, NVIDIA NemoClaw and Goose

02 · Our cloud

나가기 전에 봉인됨

$19 per node, per month

The snapshot is encrypted on the node with a key generated there and never sent. Your browser decrypts it. Subpoena us and we hand over a blob we cannot open.

  • AES-256-GCM, key held on the node
  • Decryption happens client side
  • Stop, Pause and Resume reach any node on the account

03 · Self-hosted

고객 인프라, 단일 테넌트

$190 per node, per year

One container serves the UI and the ingest API on your own network, with node tokens and admin auth. Licences verify offline against an embedded Ed25519 key you can fingerprint and compare.

  • No calls to clawmetry.com beyond an off-by-default version ping
  • Same feature set, same runtimes, same rate
  • Two months free against the monthly price

배포 방식에 따라 요금이 달라지지 않습니다. 셀프 호스팅도 클라우드 플랜과 동일하게 노드당 월 $19 또는 노드당 연 $190입니다. 저희 서버가 아니라 소프트웨어에 지불하시는 것이기 때문입니다. 달라지는 것은 데이터를 누가 보관하고 컨테이너를 누가 운영하느냐입니다. 상담 없이 셀프 호스팅 키를 바로 구매하세요.

하나의 요금, 하나의 기준

누구와 이야기하기 전에
직접 계산해 보세요.

7 days, every runtime, no card

$19per node, per month

Pro, billed monthly. Starter is $9 and covers fewer features, not fewer runtimes. Annual is $190 per node, two months free. Free forever for OpenClaw, NVIDIA NemoClaw and Goose.

Start the trial
  • Every runtime, including Claude Code, Codex and Cursor.
  • Cloud or fully self-hosted, at the same rate. Unlimited chat channels.
  • Waste flags, error triage, cost optimizer, OTLP export, tool policy.
  • Cancel any time in Stripe. No contract, no seat count.

Already decided? Buy now and install after. Your key is on the next screen.

예상 비용

대시보드를 볼 사람 수가 아니라, 데몬이 실행될 머신 수를 세십시오.

25
4

$551 $6,612 per year · 29 nodes

A node is a machine running the daemon and reporting a heartbeat. The unit of attribution is a machine, so spend maps to a laptop, not to a team. The same number applies self-hosted, so you can price a deployment on your own infrastructure off this figure without asking us. Ephemeral CI runners are not yet specified; the fleet brief carries the rule in writing before you sign.

리스크를 책임지는 분들을 위해

전사 규모로 운영하기.

지난 분기에 엔지니어들이 Claude Code, Cursor, Codex, Copilot을 설치했습니다. 아무도 정식으로 도입하지 않았지만, 전부 회사의 저장소와 키에 접근하고 있습니다.

에이전트에 설치할 것이 없습니다

Each runtime is read from the session store it already writes. No SDK, no wrapper, no second agent on the laptop, nothing a developer has to opt into or can forget.

A control layer that governs only its own SDK is blind to everything a developer runs outside it.

SSO의 해답은 셀프 호스팅입니다

SSO does not exist yet. Single-tenant self-hosted does: one container serves the UI and the ingest API, with node tokens and admin auth, and licences verify offline against an embedded Ed25519 key you can fingerprint and compare.

A self-hosted deployment makes no calls to clawmetry.com beyond an off-by-default version ping.

지금 쓰는 네트워크에 그대로 맞습니다

SIEM and syslog export in CEF or JSON over TCP, UDP or TLS, emitted daemon-side because the cloud never holds plaintext. Corporate TLS interception is supported, with a doctor command that walks DNS, TCP, proxy and TLS in order.

In: your collector feeds us over OTLP. Out: full-event SIEM export. Not yet: correlated spans carrying your trace ids.

플릿 배포는 현재 수동입니다

There is no Helm chart, no Terraform provider and no Intune or Jamf package. If a daemon on developer laptops is a non-starter, run the enforcement proxy centrally and govern spend without touching an endpoint.

Auto-update ships on by default. CLAWMETRY_AUTO_UPDATE=0 turns it off. No staged rollout, no version pinning.
Get the fleet brief The gap list Both dated and versioned. The gap list names everything we have not built, including SSO, RBAC, SOC 2 and data residency.

매번 받는 질문들.

실행 전에 차단할 수 있나요, 아니면 실행됐다고 알려주기만 하나요?

Both, and the split is in the table above. Claude Code gets a genuine pre-execution tool gate, and anything routed through the proxy gets spend refused with a 429 before tokens are spent. For the other runtimes it is detection plus process control: we can see it and end it, we cannot veto the call it already made.

에이전트를 알아서 중단시키나요?

No. Detectors write a warning and stop there. On the reference install every flagged session ran to completion. A run ends only when a person ends it, or when you have set a spend cap yourself and an agent crosses it.

에이전트 실행 방식이 바뀌나요?

Read-only by default. Two things you opt into change that: the proxy, a separate process you point one env var at, and approval gates, where ClawMetry adds a gate to the runtime’s own permission system and removes only what it added.

체험 기간이 끝나면 어떻게 되나요?

The paid runtimes stop being read and the free three keep working. Entitlements are a commercial boundary, not a technical one. We do not hard-block a running fleet on a licence check.

누가 만들고 있나요, 그리고 사라지면 어떻게 되나요?

One person, plus contributors. The core is MIT and runs on your machine, so if we vanish your install keeps working and your data stays in a DuckDB file you own. The paid layer is a separate closed package. Ask for the fleet brief and it names the legal entity, the largest deployment we run today, and the reference count, with no rounding.

보안 검토를 통과할 수 있나요?

For a pilot on non-production machines, usually: MIT source you can audit, local-first storage, single-tenant self-hosted and offline licence verification are the arguments that work. For a regulated production rollout, read the gap list first. SSO, RBAC beyond owner and member, SOC 2, CAIQ and a third-party pen test are all absent, and that is a two-to-four quarter conversation.

최신 소식을 받아보세요

새로운 기능, 릴리스, 그리고 성장하는 -metry 생태계 소식을 받아보세요.