23 docs tagged with "Observability"

Build adaptive trust for AI agents: Automatically adjust policies and permissions based on behavior patterns and changing risk levels.

Trace an AI agent from repository changes to registered runtime, governance configuration, sessions, and audit evidence.

Never miss a policy violation: Set up real-time alerts for trust changes, risky actions, and compliance issues via Slack or email.

Reconstruct why an agent misbehaved from sealed proof-trace evidence, and get an intervention-driven path to a repaired retry.

Monitor all your AI agents in one dashboard: View trust scores, active sessions, policy violations, and real-time alerts.

Understand how DeepAgents invocations, model calls, tools, subagents, and telemetry are represented in OpenBox.

Understand how LangChain agent runs, model calls, tools, and telemetry are represented in OpenBox.

Understand how LangGraph graph runs, tools, LLM calls, signals, and telemetry are represented in OpenBox.

Understand how Mastra workflows, tools, agents, and telemetry are represented in OpenBox.

Understand how n8n agent runs, model calls, tools, and telemetry are represented in OpenBox.

Org-wide registry of agents, models, tools, and integrations, including unregistered callers OpenBox has observed but you haven't registered yet.

Catch AI agent issues before they escalate: Real-time anomaly detection, policy violation alerts, and complete execution audit trails.

Visualize agent activity, handoffs, and governance verdicts for a team's multi-agent runs in one timeline.

Understand the trust lifecycle: The 5-stage process for governing AI agents from identity verification to continuous adaptation.

Connect repositories to OpenBox Projects and trace agents from code changes to governed runtimes, sessions, and governance snapshots.

Replay any AI agent session step-by-step: Audit every decision, investigate policy violations, and reconstruct complete execution history.

OpenTelemetry capture behavior for the OpenBox Deep Agents SDK: model calls, tools, HTTP, file I/O, database operations, and DID-signed requests.

See what operational telemetry the OpenBox LangChain SDK captures and how it appears in OpenBox.

See what operational telemetry the OpenBox LangGraph SDK captures and how it appears in OpenBox.

See what operational telemetry the OpenBox Mastra SDK captures and how it appears in OpenBox.

See what operational telemetry the OpenBox n8n node captures and how it appears in OpenBox.

What counts as a trust incident in OpenBox: HALT decisions, critical behavioral violations, trust-tier demotions, and drift crossing into a critical band.

See your organization's trust health: Aggregate trust scores, compliance rates, and risk trends across all AI agents in one view.