Monitor activity

Every Observe task in Control Center: trace runs, read analytics, audit changes, inspect logs, and export or report on activity.

This guide is organized by task. Find what you want to do in the list below, then follow the steps. Each task links to the reference page it comes from.

Everything in Control Center is account-wide; most Observe screens can be narrowed to one project. What you can see depends on your Control Center role.

Task	Go to
Find an agent or app run	Find a run
Inspect what happened during a run	Inspect a trace
Understand a run's status	Run status
See account-wide build activity	Overview
See how published apps are used	Apps
See how agents perform	Agents
See who changed what in the account	Audit Trails
See every model call, its tokens and cost	Gateway Logs
See every embedding call	Embedding Logs
See requests the platform received	Incoming requests
See requests the platform sent out	Outgoing requests
See database operations	DB Query Logs
Export activity data, once or on a schedule	Export
Change or re-download an export	Track exports
Review a usage or execution report	Reports
Email a report	Email a report

---

Runs & Traces

Runs & Traces records every agent and app execution in the account: LLM calls, tool calls, guardrails, inputs, and outputs at each step. It is read-only: you can't rerun or edit executions here.

Find an agent or app run

1

Open Runs & Traces

Open Observe › Runs & Traces. Runs appear newest first, with 10 threads on each page.

2

Narrow the results

Combine any of these; they combine with AND:

• Project — runs from one project.
• Agent name or ID — runs from one agent.
• Thread or execution ID — one run by its full identifier.
• Business ID — runs linked to a business identifier.
• Date range — runs by start time.

3

Open the thread

Open a thread to see its agent name, full thread ID, and traces. Each trace shows its opening prompt, ID, start time, duration, and span total.

Runs & Traces.

IDs may appear shortened; point to one to reveal the full value. Refresh the results without changing the URL. Reference: Runs & Traces › Find an execution.

Inspect what happened during a run

Open a trace, then choose a view:

• Trace — spans in execution order, with their input and output. Open a span to see its name, timestamp, duration, input, and output as JSON. LLM spans also report the model, provider, input and output tokens, credits, and USD cost.
• Graph — parent and child span relationships, each tagged with its category: Run, LLM, Tool, Retrieval, Reasoning, Guardrail, or Privacy.
• Summary — a plain-language walkthrough with duration, span, and token totals.

Copy the full trace ID into a ticket, log query, or API call. No traces to display means the logs couldn't reconstruct spans; it doesn't mean the run failed. Reference: Runs & Traces › Inspect a trace.

Understand a run status

• Completed — the run finished normally.
• Waiting — the run is executing, or paused on an external step, which can include an item awaiting Human Review. The status doesn't distinguish these.

Waiting doesn't automatically mean a run is stuck. If it exceeds its usual duration, check Human Review before treating it as a failure. Reference: Runs & Traces › Status values.

---

Analytics

Analytics is one page with three tabs. Filter by project; each tab has its own date-range filter. Open Observe › Analytics.

See account-wide build activity

On the Overview tab (the default):

• Platform Events — Active Users and Projects Created, with an Acquisition funnel.
• App Events — Apps Created, Build Started, and Apps Published, with an App build funnel.
• Agent Events — Build Started, Build Completed, Runs Initiated, and Agent Mapped.

Where a stage has records behind it, View records opens the underlying list.

Analytics, Overview tab.

See how published apps are used

On the Apps tab, narrow by project or date range, search for an app, or reload. It reports Unique Visitors, Total Page Views, and Published Apps; unique visitors over time; browsers; how visitors split across apps; and each published app's Visitors and Pageviews, under the subdomain it was published to.

See how agents perform

On the Agents tab, narrow by project, agent, or date range. It reports Total Executions, Success Rate, Avg Duration, and Active Agents; Executions Over Time; Execution States (Completed / Failed / Aborted); Quality & Learning › Eval Pass Rate; and an Agents table with each agent's Executions, Success, and Avg Duration.

The Agents tab needs agent executor version 1.4 or above. On an older executor, it shows a banner instead of data.

Reference: Analytics.

---

Audit and logs

See who changed what in the account

Open Observe › Audit Trails. Filter by project, actor, or date range, and search by actor, module, or action. Entries appear most recent first, each with its Timestamp, Actor (by email), Module (such as Accounts, Users, Knowledge, or Agents), and Action (such as Created, Updated, or User Invited).

Audit Trails.

Reference: Audit Trails.

See every model call

Open Observe › LLM Logs, on the Gateway Logs tab. Narrow by project and model, and search by model or agent. Each row is one completed call, with its Model, For (the target), Tokens, Cost, Latency (ms), When, and Status. Select a row for the full detail.

LLM Logs, Gateway Logs tab.

See every embedding call

On LLM Logs, select Embedding Logs. Every query sent to an embedding model is listed, including failed calls. Narrow by project, model, and date range; search by query or model. Each row shows the Query, Model, When, and Status (Success or Failed). Expand a row for the full query text, the model, and the duration in milliseconds.

Reference: LLM Logs.

See requests the platform received

Open Observe › Request Logs, then Incoming Request Logs. Narrow by project and date range, filter by module (Agent, App, Skill, Integrations, Marketplace, Platform) and method (GET, POST, PUT, PATCH, DELETE), and search the log. Each row shows the Method, URL, and When. Select a row for the full request.

See requests the platform sent

Open Observe › Request Logs, then Outgoing Request Logs. The same filters apply. Each row shows the Method (CUSTOM for calls that don't map to a standard verb), Status, URL, Response Time (ms), and When.

Request Logs don't record traffic that doesn't pass through Qore. Reference: Request Logs.

See database operations

Open Observe › DB Query Logs. Narrow by project and module, and search the log. Each row shows the Module, the Operation in schema.table.action form (such as agents.agent.update), the User, and the Timestamp.

DB Query Logs.

Reference: DB Query Logs.

---

Export and reports

Export activity data

1

Open Export

Open Observe › Export and start a new export. Exports run in the background.

2

Name and scope it

Give it a name, such as Weekly audit trail export. Choose every project (All projects, the default) or one project.

3

Choose the datasets

Select at least one: Credit Usage, Platform Analytics, App Analytics, Agent Analytics, Audit Trails, AI Gateway Logs, or Eval Results. Each dataset becomes its own file.

4

Choose the format and delivery

• Format — CSV (default), JSON, or PDF.
• Email (default) — comma-separated recipients; leave it empty to send to your signed-in account.
• Webhook — an HTTPS URL that receives a POST with the download link, and an optional signing secret.

5

Choose when it runs

• Scheduled (default) — hourly, daily, weekly, or monthly. Daily is the default, at 02:00 UTC; weekly runs start on Monday, monthly on the first.
• One-time — starts immediately, for Today, Yesterday, Last 7 days, Last 30 days, This Week, This Month, or Last Month.

6

Confirm

Review the summary, such as Weekly at 08:00 UTC · All projects · 1 dataset · Csv → Email, then save or run it.

All times are UTC. Reference: Export.

Track and update exports

• Exports tab — saved definitions, with their project, datasets, mode, format, delivery, next run, last run, and status. Update a definition to change future runs.
• History tab — completed runs, with the requested time, project, datasets, delivery, format, row total, and status.

If an email or webhook doesn't arrive, retrieve the file from the history while its link is still valid.

Export, History tab.

Review a usage or execution report

Open Observe › Reports. There are two fixed reports:

• Credit Usage Report — spend and activity per project: apps billed, agent runs, and their cost.
• Agent Executions Report — each agent's run count, average cost, and average duration.

For a report, choose the project scope (All projects by default), the date range (This Month by default, or a preset), and generate it. Results appear as a table.

Reports.

Email a report

Choose the project scope, date range, and delivery format (PDF, the default, CSV, or JSON), then request the email. The file goes to the signed-in user when it's ready. Reports can't be downloaded from the page or run on a schedule; for scheduled files, use Export.

Reference: Reports.

---

Related pages

• Observe and Control Center
• Costs: Manage costs and limits. Quality: Evaluate agent quality.

---

Maintained by Abhishek Paul AI-assisted, human-approved