Skip to content

GH-100 · Domain 5: Monitor and optimize GitHub usage · 13% of the exam

Task 5.1: Monitor enterprise usage and activity

Searching and querying the audit log and API activity, and reading usage and adoption data to find inactive users and features nobody uses.

Study it

  • Audit log queries and API activity

    Lesson coming

  • Tracking adoption against cost: usage metrics, dormant users and features nobody uses

    Partly covered by: Measuring AI Value

Sample questions

Try each one before opening the answer. Every option is explained, with the GitHub documentation page that proves it.

Question 1 · choose 1

On 14 November 2026, an auditor asks for every change between the member and owner roles in the enterprise's organizations since 1 June 2026. The enterprise owner searches the enterprise audit log for action:org.update_member, and the oldest result is from mid-August. Which query returns the events she needs?

  1. ANone: events before mid-August are gone, so streaming should be set up for future audits
  2. Baction:org.update_member operation:modify
  3. CThe text "promoted to owner" searched across the audit log
  4. Daction:org.update_member created:>=2026-06-01
Show the answer and why
  • ANone: events before mid-August are gone, so streaming should be set up for future audits

    Incorrect

    The audit log keeps events from the last 180 days; only the default view is limited to three months.

  • Baction:org.update_member operation:modify

    Incorrect

    The operation qualifier narrows the results; it doesn't reach past the default three months.

  • CThe text "promoted to owner" searched across the audit log

    Incorrect

    The audit log can't be searched with text; queries use key and value filters.

  • Daction:org.update_member created:>=2026-06-01

    Correct

    Events older than three months are shown only when a date range is given with the created qualifier.

The enterprise audit log lists events from the last 180 days, but by default only the past three months are displayed; older events need a date range with created. Queries are built from key:value filters such as action, actor, created and operation, and free text isn't searchable. The org.update_member event records a role change between owner and member.

Question 2 · choose 1

Finance proposes removing everyone in the enterprise's dormant users report from their organizations to cut license costs. The report lists, among others, engineers who push to private repositories over SSH every day and rarely open the web UI. What should the enterprise owner tell finance?

  1. AReview the list first: Git operations on private repositories don't count as activity, and inactive members do consume licenses
  2. BRemove everyone listed, because dormant users have had no Git or web activity in 30 days
  3. CRemove nobody, because dormant users don't consume a license and nothing would be saved
  4. DMark the SSH engineers as active in the report, then remove everyone who is still listed
Show the answer and why
  • AReview the list first: Git operations on private repositories don't count as activity, and inactive members do consume licenses

    Correct

    Daily SSH pushers can appear dormant, while removing truly inactive members does lower license use.

  • BRemove everyone listed, because dormant users have had no Git or web activity in 30 days

    Incorrect

    Git operations on private repositories and access with SSH keys aren't counted as activity.

  • CRemove nobody, because dormant users don't consume a license and nothing would be saved

    Incorrect

    Dormant users who are members of an organization in the enterprise consume a license.

  • DMark the SSH engineers as active in the report, then remove everyone who is still listed

    Incorrect

    A dormant user can't be marked as active; the user has to perform one of the counted activities.

A user is dormant after 30 days without any of the activities GitHub counts. Accessing resources with a personal access token, SSH key or GitHub App, and Git operations on private repositories, aren't counted, so active engineers can appear in the report. Dormant users who belong to an organization in the enterprise still consume a license.

Question 3 · choose 1

On 14 November 2026, an investigator suspects that the service account svc-build has cloned repositories it shouldn't since about 1 November. She calls GET /enterprises/acme/audit-log with phrase=actor:svc-build+created:>=2026-11-01 and gets only events such as sign-ins and settings changes, no clones. Which call returns the clone events she can still get, and what period do they cover?

  1. AThe same call with include=git, which returns svc-build's clone events from 1 November onward
  2. BThe same call with action:git.clone added to the phrase, which returns clone events from the last seven days
  3. CThe same call with include=git, which returns clone events from the last seven days only
  4. DThe same call with include=all, which returns web and Git events back to 1 November
Show the answer and why
  • AThe same call with include=git, which returns svc-build's clone events from 1 November onward

    Incorrect

    The audit log retains Git events for seven days, so clones from before 7 November are gone.

  • BThe same call with action:git.clone added to the phrase, which returns clone events from the last seven days

    Incorrect

    The endpoint returns web events by default; Git events need include=git or include=all.

  • CThe same call with include=git, which returns clone events from the last seven days only

    Correct

    include=git adds Git events to the results, and the audit log keeps them for seven days.

  • DThe same call with include=all, which returns web and Git events back to 1 November

    Incorrect

    Git events are kept for seven days, while the 180-day retention applies to web events.

The enterprise audit log REST endpoint returns web (non-Git) events by default; include=git returns Git events such as clones, fetches and pushes, and include=all returns both. The audit log lists events from the last 180 days but retains Git events for only seven days, so on 14 November the clone events reach back to 7 November. Keeping Git events longer requires streaming them before they age out.

Question 4 · choose 1

An engineering director wants raw daily Copilot activity per team, for the 40 teams of an organization, loaded into the company's business intelligence tool. Which approach does GitHub document?

  1. ACall a team-level Copilot metrics endpoint once a day for each team and load the results directly
  2. BJoin the daily user-teams report with the daily per-user usage report, then aggregate by team
  3. CDownload the enterprise's dormant users report and filter it by team
  4. DExport the team view of the Copilot impact dashboard as raw data
Show the answer and why
  • ACall a team-level Copilot metrics endpoint once a day for each team and load the results directly

    Incorrect

    The Copilot usage metrics APIs don't publish a single pre-aggregated team report.

  • BJoin the daily user-teams report with the daily per-user usage report, then aggregate by team

    Correct

    Team-level metrics are built by joining the user-teams report with the per-user usage report.

  • CDownload the enterprise's dormant users report and filter it by team

    Incorrect

    The dormant users report covers activity on GitHub, not Copilot usage.

  • DExport the team view of the Copilot impact dashboard as raw data

    Incorrect

    The dashboard gives a UI view of team adoption; raw per-team data comes from joining the reports.

The Copilot usage metrics APIs don't provide a team report. Team-level metrics are constructed by joining the daily user-teams report, which lists each user's team memberships, with the daily per-user usage metrics report, then aggregating the joined rows by team. The impact dashboard shows team adoption in the UI without that join.

Question 5 · choose 1

An enterprise owner has the SHA-256 hash of a compromised token and wants every audit log event that token performed. Which search should she enter in the audit log on GitHub?

  1. Atoken:VALUE
  2. Bhashed_token:"VALUE"
  3. Cactor:VALUE
  4. Dhashed_token:VALUE without quotation marks
Show the answer and why
  • Atoken:VALUE

    Incorrect

    The qualifier for token searches is hashed_token.

  • Bhashed_token:"VALUE"

    Correct

    Searches by token use hashed_token with the hash wrapped in quotation marks.

  • Cactor:VALUE

    Incorrect

    actor takes the name of the account that acted, not a token hash.

  • Dhashed_token:VALUE without quotation marks

    Incorrect

    The hashed token value must be wrapped in quotation marks.

To find events associated with a token, include hashed_token:"VALUE" in the audit log search, with the token's SHA-256 hash wrapped in quotation marks. With the REST API, the hash must also be URI-escaped. Searches don't return Git events.

Question 6 · choose 1

Users report slow pages on a GitHub Enterprise Server 3.22 instance. Before opening a support ticket, the site administrator wants to look at historical CPU and storage use and response times. Where should she look?

  1. AThe Monitor dashboards in the Management Console
  2. BThe enterprise audit log, filtered to the instance's admin actions
  3. CThe GitHub Status page
  4. DThe instance's license usage report
Show the answer and why
  • AThe Monitor dashboards in the Management Console

    Correct

    The monitor dashboards show historical data such as CPU, storage and response times.

  • BThe enterprise audit log, filtered to the instance's admin actions

    Incorrect

    The audit log records user and admin actions, not system metrics.

  • CThe GitHub Status page

    Incorrect

    The status page covers GitHub's own services, not a self-hosted instance.

  • DThe instance's license usage report

    Incorrect

    License reports show seats, not system performance.

The monitor dashboards, opened from Monitor in the Management Console, visualize historical data such as CPU and storage use, application and authentication response times, and general system health. The Operational Health dashboard is shown by default.

Question 7 · choose 1

An organization owner compares the GitHub Actions usage metrics with the Actions charges on the bill and finds that the minutes don't line up, mostly for macOS jobs. What explains this?

  1. AUsage metrics include only self-hosted runners
  2. BUsage metrics are updated once a month
  3. CUsage metrics don't apply minute multipliers
  4. DUsage metrics exclude private repositories
Show the answer and why
  • AUsage metrics include only self-hosted runners

    Incorrect

    Usage metrics compare self-hosted and GitHub-hosted runners.

  • BUsage metrics are updated once a month

    Incorrect

    The difference comes from multipliers, not refresh timing.

  • CUsage metrics don't apply minute multipliers

    Correct

    They show where minutes are used, not billed amounts with multipliers.

  • DUsage metrics exclude private repositories

    Incorrect

    Usage metrics cover the organization's repositories.

GitHub Actions usage metrics show how many minutes workflows and jobs consume, by workflow, job, repository, runtime OS and runner type. They don't apply minute multipliers, so their primary purpose is to show how and where minutes are used rather than to match the bill.

Practice domain 5 →Practice all domains →