1.1 Exam Structure, Format & Scoring
Key Takeaways
- Microsoft Learn allots 100 minutes to complete the GH-200 assessment; the exam is proctored and may include interactive components.
- GH-200 is scored on a scale of 1 to 1,000 with a passing score of 700; Microsoft states that because this is a scaled score it may not equal 70% of the points.
- Neither GitHub nor Microsoft publishes an item count for GH-200; Microsoft confirms only that some questions are unscored and that candidates cannot tell which ones.
- The current skills-measured list is dated January 2026 and has five domains: Author and manage workflows (20-25%), Consume and troubleshoot workflows (15-20%), Author and maintain actions (15-20%), Manage GitHub Actions for the enterprise (20-25%), and Secure and optimize automation (10-15%).
- The retake ladder is 24 hours after a first failure, then a 14-day wait between all subsequent attempts, with a maximum of five attempts in the 12 months following your first attempt.
Exam Structure, Format & Scoring
The GitHub Actions certification (exam code GH-200) validates your ability to automate software development workflows with GitHub Actions: authoring and maintaining workflows and actions, managing Actions at scale for an organization or enterprise, and keeping that automation secure and efficient.
Since July 1, 2025, GitHub certifications are delivered through Pearson VUE and appear in your Microsoft Learn profile. Older exam guides that describe a PSI-delivered GitHub certification program with different item counts and retake rules describe a retired program. Everything in this section is taken from the currently published Microsoft Learn certification page and the GH-200 study guide.
1. Exam Logistics & Administration
| Specification | Currently published value | What it means for you |
|---|---|---|
| Exam code | GH-200 | The credential is listed on Microsoft Learn as "GitHub Actions". |
| Level | Intermediate | Aimed at DevOps engineers, developers, and IT pros already using Actions. |
| Time allowed | 100 minutes | Microsoft Learn states: "You will have 100 minutes to complete this assessment." |
| Number of items | Not published | Microsoft does not disclose the item count for GH-200. Do not plan against a fixed number. |
| Passing score | 700 on a 1-1,000 scale | Microsoft: "As this is a scaled score, it may not equal 70% of the points." |
| Delivery | Pearson VUE test center or online proctored | Proctored; the exam "may have interactive components". |
| Scheduling | Through Microsoft Learn; Pearson VUE books up to 90 days ahead | Register with a personal Microsoft account (MSA), not a work/school account. |
| Languages | English, Spanish, Portuguese (Brazil), Korean, Japanese | If the exam is not offered in your language, you may request 30 extra minutes. |
| Price | $99 USD in most regions | Microsoft prices the exam "based on the country or region in which the exam is proctored". |
| Prerequisites | None | Intermediate hands-on Actions experience is assumed but not required. |
| Retake policy | 24 hours after the 1st failure; 14 days between all subsequent attempts; max 5 attempts per 12 months from the first attempt | You pay again for each attempt. |
| Skills measured as of | January 2026 | The exam "changed significantly" in January 2026 - objectives were added, removed, regrouped, and reworded. |
[!IMPORTANT] Do not study against invented numbers. Community posts and third-party courses still repeat a "75 questions - 60 scored plus 15 pretest" figure. That figure comes from the retired PSI-era program; it is not published for the current Pearson VUE exam. Microsoft publishes the 100-minute window and the 700/1,000 cut score, and nothing else about exam length.
Test Center vs. Online Proctoring
- Pearson VUE test center: standardized hardware, staffed check-in, no dependence on your home network. Bring a valid, unexpired, government-issued photo ID whose first and last name match your registration exactly.
- Online proctored (OnVUE): run the system pre-check before registering. You must sit alone in a private room with a cleared desk and uninterrupted webcam and microphone feeds. Another person entering the room, leaving the camera frame, or opening unauthorized applications can end the session.
[!TIP] Try the sandbox first. Microsoft publishes a free GitHub exam sandbox at
https://GHCertDemo.starttest.comthat reproduces the exam user interface and question types. There is also a free practice assessment on the GH-200 certification page. Use both before exam day - they remove all surprise from the interface.
2. Scoring Mechanics & Unscored Items
GH-200 uses scaled scoring. Microsoft converts your raw performance into a standardized score between 1 and 1,000, and 700 or greater passes.
+-----------------------------------------------------------------------------+
| GH-200 SCALED SCORING MODEL |
| |
| [1 MINIMUM] --------------- [700 PASSING SCORE] --------------- [1000 MAX]|
| | | | |
| v v v |
| Failing range Minimum demonstrated Strong |
| (see the skill bar chart) competence performance |
+-----------------------------------------------------------------------------+
Four scoring rules matter on exam day, and all four come straight from Microsoft's exam scoring documentation:
- 700 is not 70%. The cut score is fixed on the scale, not on the raw item count. Harder question sets require fewer raw points to reach 700; easier sets require more. This is why Microsoft never publishes a "percent correct" target.
- There is no penalty for guessing. An incorrect answer simply earns no point. An unanswered question earns no point either - so never leave one blank.
- Some items are unscored. Microsoft seeds new questions to gather statistics before they count. You cannot tell which ones they are, so answer every item as if it will be scored.
- Multi-part items can be scored per component. On most multi-part questions you earn one point for each correctly answered component, so partial credit is possible. Where a question is worth more than one point, the question text says so.
Your score report is available within minutes of finishing (about 30 minutes if the exam contains labs) and includes your numeric score, pass/fail status, and a bar chart of performance by skill area. The bar chart shows relative strength and weakness only - you cannot back-calculate how many questions you answered correctly from it.
3. Official Blueprint: Skills Measured as of January 2026
The exam covers five skill areas. Use the official names and weight bands below; several third-party sites still publish a retired four-domain split (40/20/25/15) that no longer matches the exam.
+-----------------------------------------------------------------------------+
| GH-200 SKILLS AT A GLANCE (JANUARY 2026) |
| |
| Author and manage workflows ===========> 20-25% |
| Consume and troubleshoot workflows ========> 15-20% |
| Author and maintain actions ========> 15-20% |
| Manage GitHub Actions for the enterprise ===========> 20-25% |
| Secure and optimize automation =====> 10-15% |
+-----------------------------------------------------------------------------+
| Official domain | Weight | Representative objectives |
|---|---|---|
| Author and manage workflows | 20-25% | Triggers for scheduled, manual, webhook, and repository events; workflow_dispatch inputs and workflow_call input/secret mapping; jobs, steps, conditional logic, and needs; workflow commands and environment variables; service containers; strategy/matrix with include/exclude, fail-fast, max-parallel, and runner-image changes; YAML anchors, aliases, and merge keys; predefined contexts and immutable-action pinning; ${{ }} expression evaluation; editor tooling; caching, artifacts, and retention via REST API; GITHUB_ENV/GITHUB_OUTPUT; GITHUB_STEP_SUMMARY job summaries; status badges and environment protections. |
| Consume and troubleshoot workflows | 15-20% | Reading triggers and effects from configuration and logs; diagnosing failed runs; expanding YAML anchors and merged mappings when analyzing a workflow; interpreting matrix expansions and re-running individual matrix jobs; locating logs and artifacts in the UI and API; consuming organization-level, reusable, and non-public organization workflow templates; starter workflows vs. reusable workflows vs. composite actions; contrasting disabling and deleting workflows. |
| Author and maintain actions | 15-20% | JavaScript, Docker, and composite action types plus immutable-actions implications; troubleshooting action execution; required files, directory structure, and action.yml metadata; workflow commands inside actions; public/private/Marketplace distribution; Marketplace publishing; versioning and release strategy. |
| Manage GitHub Actions for the enterprise | 20-25% | Defining and governing reusable components and templates; controlling access to actions and workflows; organization use policies; configuring and monitoring GitHub-hosted and self-hosted runners; IP allow lists and networking; runner groups and runner troubleshooting; preinstalled software, the toolcache, and installing tools at runtime; scoping encrypted secrets and variables at organization, repository, and environment level, including REST API management. |
| Secure and optimize automation | 10-15% | Environment protections and approval gates; identifying trustworthy Marketplace actions; script-injection mitigation; GITHUB_TOKEN lifecycle, granular permissions, and contrast with a PAT; OIDC federation via id-token; full-commit-SHA pinning; organization allow/deny lists and required reviewers for unverified actions; artifact attestations and provenance; caching and retention tuning; scaling and optimization strategy. |
[!NOTE] GA versus preview. Microsoft states that most questions cover generally available features, but the exam "may contain questions on Preview features if those features are commonly used". Do not skip a widely adopted preview capability just because it is labeled preview.
4. Question Formats
Microsoft does not publish a format breakdown for GH-200, but the certification page confirms the exam is proctored and "may have interactive components", and the sandbox demonstrates the item types you will meet.
+-----------------------------------------------------------------------------+
| GH-200 ITEM TYPES YOU WILL MEET |
| |
| [1. SINGLE-SELECTION] ----> One correct answer |
| (Syntax precision, keyword behavior) |
| |
| [2. MULTIPLE-SELECT] ----> The stem states how many to choose |
| (e.g., "Select TWO", scored per component) |
| |
| [3. SCENARIO / YAML] ----> Workflow snippet, log excerpt, or policy |
| (Find the flaw, predict the outcome, fix it) |
+-----------------------------------------------------------------------------+
The heaviest cognitive load on GH-200 is scenario analysis over YAML: you are shown a workflow file, a run log, or an enterprise policy configuration and asked to explain a behavior, locate a defect, or choose the compliant design. That is the skill this guide drills chapter by chapter.
5. Time Management Across the 100-Minute Window
Because the item count is not published, plan your pacing in fractions of the window, not in seconds per question. Do a quick arithmetic check in the first minute: divide 100 minutes by the number of items the interface reports, and use that as your per-item ceiling.
+-----------------------------------------------------------------------------+
| 100-MINUTE THREE-PASS PACING MODEL |
| |
| +---------------------------------------------------------------------+ |
| | PASS 1: FIRST SWEEP (first ~65% of the clock, ~65 minutes) | |
| | - Answer every item once; never leave one blank. | |
| | - Move fast on recall items; give a best guess on long scenarios. | |
| | - Flag anything that takes more than your per-item ceiling. | |
| +---------------------------------------------------------------------+ |
| | |
| v |
| +---------------------------------------------------------------------+ |
| | PASS 2: FLAGGED SCENARIO DEEP-DIVE (~25% of the clock, ~25 minutes) | |
| | - Trace YAML dependency graphs, matrix expansions, and contexts. | |
| | - Re-read multi-select stems to confirm the required answer count. | |
| +---------------------------------------------------------------------+ |
| | |
| v |
| +---------------------------------------------------------------------+ |
| | PASS 3: INTEGRITY CHECK (final ~10% of the clock, ~10 minutes) | |
| | - Confirm zero unanswered items remain. | |
| | - Resist rewriting answers you have no new reason to doubt. | |
| +---------------------------------------------------------------------+ |
+-----------------------------------------------------------------------------+
[!TIP] Pacing checkpoints in percentages: at the 1/3 mark you should have seen a third of the items; at the 2/3 mark, two thirds. If you are behind at either checkpoint, start guessing-and-flagging aggressively - an unanswered item is worth exactly as much as a wrong one, and you cannot earn points on questions you never reach.
6. After the Exam
- Renewal: the GH-200 certification page does not publish an expiration date for this credential. The GH-200 study guide links to Microsoft's certification renewal policy, under which associate, expert, and specialty certifications expire annually and are renewed free through an unproctored online assessment on Microsoft Learn during a six-month eligibility window. Check the expiration shown on your own Microsoft Learn profile after you pass rather than relying on third-party claims of a two-year or three-year validity - published sources disagree.
- If you fail: review the skill-area bar chart on your score report, study the weakest areas and the heaviest-weighted domains, and rebook after the 24-hour wait. Microsoft does not disclose which individual questions you missed.
A candidate has 100 minutes for the GH-200 exam and hits an unusually intricate enterprise-policy scenario roughly a third of the way through the first pass. How should they handle it, given how GH-200 is scored?
According to the GH-200 skills-measured list dated January 2026, which two skill areas carry the highest published weight bands?
A candidate fails GH-200 on their first attempt, then fails again on the retake. When may they sit the exam a third time, and how many attempts are they allowed in total?