Open-source, OTel-native observability for autonomous AI agents. Now available on PyPI and npm. MIT licensed.
FEATURES
INTEGRATIONS
Every other observability tool is built for LLM developers. OpenClawWatch is built for people whose agents have real-world side effects.
Agents running overnight have full access to email, files, and services. Without observability, you find out what happened when a customer complains.
An agent looping on a failing tool call can burn $40/hr. Without per-agent cost tracking, you get the bill at the end of the month.
Behavioral drift detection, safety alerts, schema validation — they all require API keys, hosted backends, and credit cards. OpenClawWatch runs on your machine.
Full GenAI Semantic Conventions compliance from day one. Agent spans, tool calls, token metrics — exportable to Grafana, Jaeger, Datadog, or any OTel backend without transformation.
OTel SemConv v1.37+Real-time USD cost per LLM call, attributed to the agent and tool that triggered it. Configurable daily/session/per-agent budget alerts fire before you get the bill.
per-model pricing TOMLThe only observability tool built for agents with real-world side effects. Configurable alerts fire on email sends, file writes, form submissions, and payment actions.
unique to OpenClawWatchDeterministic, no-cloud drift detection. Automatically baselines token usage, tool call sequences, output schema, and session duration — alerts when agents deviate.
no API key requiredJSON Schema validation for tool outputs and agent responses. Declare schemas per-agent/tool in config, or use inference mode to auto-derive from observed sessions.
JSON Schema draft-07A full-featured CLI (ocw status / traces / cost / drift) with JSON output on every command. Local API at localhost with Prometheus /metrics endpoint, OpenAPI spec included.
pipe-friendly · scriptableWorks with every major agent runtime
The tools your team already uses are built for LLM developers. OpenClawWatch fills the gap they all leave open.
| Feature | OCW OSS | Langfuse | LangSmith | Vex | Guardrails AI |
|---|---|---|---|---|---|
| Observability | |||||
| OTel GenAI SemConv nativecompliant from day one | ✓ | ~ | ~ | ✓ | — |
| LLM call tracing | ✓ | ✓ | ✓ | ✓ | — |
| Token & cost tracking | ✓ | ✓ | ✓ | ✓ | — |
| Framework agnostic | ✓ | ✓ | — | ✓ | ✓ |
| Autonomous agent safety | |||||
| Sensitive action alertsemail, file write, payment, form submit | ✓ | — | — | — | — |
| Cost budget alertsdaily / session / per-agent | ✓ | — | — | — | — |
| NemoClaw sandbox events | ✓ | — | — | — | — |
| Retry loop detection | ✓ | — | — | — | — |
| Runtime verification | |||||
| Behavioral drift detection | ✓ | — | — | API key | — |
| Output schema validation | ✓ | — | — | ✓ | ✓ |
| Developer experience | |||||
| Fully local, no signup | ✓ | ✓ | — | — | ✓ |
| CLI interface | ✓ | — | — | — | — |
| OTLP export to any backendGrafana, Jaeger, Datadog… | ✓ | ✓ | — | — | — |
| Open source / self-hostable | ✓ | ✓ | — | ✓ | ✓ |
The commercial tier isn't a restriction — it's for teams. Multi-agent aggregation, dashboards, and SSO require cloud infrastructure that is out of scope for a local developer tool by design.
| Capability | OpenClawWatch OSS | ClawWatch Commercial |
|---|---|---|
| Telemetry collection | ✓ All runtimes | ✓ Via OSS layer |
| OTLP export (Grafana, Jaeger, etc.) | ✓ Full export | ✓ Plus hosted ingestion |
| Token & cost tracking | ✓ Single agent | ✓ Multi-agent + org budgets |
| Local safety alerts | ✓ stdout / webhook / file | ✓ Plus cloud channels |
| Behavioral drift detection | ✓ Deterministic, local | ✓ Plus LLM-as-judge |
| Output schema validation | ✓ JSON Schema | ✓ Plus custom eval rules |
| CLI interface | ✓ Full featured | — |
| Local REST API | ✓ localhost | ✓ Cloud-hosted |
| Web dashboard & charts | — intentionally excluded | ✓ |
| Multi-agent aggregation | — out of scope | ✓ |
| SSO / RBAC | — | ✓ |
| LLM-as-judge evaluation | — | ✓ |
| Hosted retention & search | — | ✓ |
Want the commercial tier? → cla.watch