ClawTrace

About ClawTrace

ClawTrace is an observability tool for OpenClaw agents that records complete execution trajectories, including every LLM call, tool use, sub-agent delegation, and associated cost. It offers both an open-source distribution and a SaaS option so teams can run it on-premises or use a managed service.

Review

ClawTrace addresses a common blind spot for agent developers by making execution history queryable and actionable, so teams can see what failed, what wasted resources, and where bottlenecks occurred. It surfaces actionable details quickly via three complementary views (execution path, call graph, and timeline) and includes a built-in "doctor" agent that can query runs and suggest fixes.

Key Features

• Complete trajectory capture: records each LLM call, tool invocation, sub-agent handoff, and cost at span level.
• Three inspection views: execution path, call graph, and timeline for different diagnostic perspectives.
• Live query agent: a built-in assistant that can analyze a run and provide recommendations for fixes and optimization.
• Fine-grained cost and performance attribution: per-span and per-sub-agent metrics with hierarchical aggregation.
• Flexible deployment: Apache 2.0 open-source repo for self-hosting plus a SOC 2 managed SaaS option for teams that prefer it.

Pricing and Value

ClawTrace is available as free open-source software under the Apache 2.0 license, which enables on-prem or hybrid deployments for teams that need full control. For teams that want a hosted option, a SaaS-managed service is offered with SOC 2 considerations for enterprise use. The core value is reduced debugging time and clearer cost attribution: by surfacing token waste, sequential tool inefficiencies, and slow LLM calls, the tool can help lower running costs and speed up agent iteration.

Pros

• Pinpoints bottlenecks and wasted tokens quickly, which helps reduce runtime cost and improve performance.
• High granularity (per-LLM-call and per-sub-agent) with hierarchical aggregation makes root-cause analysis practical.
• Offers both open-source code for self-hosting and a managed SaaS path for teams that prefer less operational work.
• Multiple views plus a live query assistant enable faster, more focused investigations of failed or costly runs.

Cons

• Some advanced automation for applying fixes is still experimental and may require human oversight during rollout.
• Self-hosted deployments that rely on a graph lakehouse architecture can add operational complexity for small teams.
• User interface and branding could benefit from polish to improve first-time user clarity.

ClawTrace is best suited for developer teams building agentic systems, platform engineers who need cost and performance visibility, and organizations that want the option to run observability tooling on-premise. It particularly helps when agents are expected to evolve over time and teams need reliable signals to guide automated or human-in-the-loop improvements.

Open 'ClawTrace' Website