3 - Manage memory, state, and execution (GH-600)

What primary limitation of 'stateless AI' does GitHub Copilot Memory aim to address?. The inability to generate code in multiple programming languages simultaneously. Multi-language support is a standard feature of. The requirement for users to repeatedly explain coding conventions and project details in prompts. The restriction on the number of files a user can have open while using the Copilot agent. The slow processing speed when analyzing large codebases for the first time.

Which specific subscription plans currently support the use of user-level preferences in Copilot Memory?. All Copilot plans, including the Free version. Only users with Copilot Enterprise subscriptions. Copilot Pro and Copilot Pro+ only. Copilot Enterprise and Copilot Business only.

If a repository-level fact or user preference is not utilized or validated by Copilot, after how many days is it automatically deleted?. 14 days. 7 days. 90 days. 28 days.

Which of the following is a limitation specific to the Copilot code review feature regarding memory usage?. It uses repository-level facts only and does not apply user-level preferences. It can only access user-level preferences and ignores repository-level facts. It requires the user to manually approve every memory fact before it is applied. It cannot access any stored memories and relies solely on the current file content.

What is the default state of Copilot Memory for users on Enterprise or Organization-managed subscriptions?. It is permanently enabled and cannot be disabled due to security protocols. It is off by default and must be enabled by an administrator. It is enabled by default for all members. It is only enabled if the user has a GitHub Pro account linked to their Enterprise seat.

How does Copilot ensure that repository-level facts remain accurate as the codebase evolves?. By only storing facts found in the README.md file, which is assumed to be updated. By requiring a manual weekly re-validation from the repository owner. By checking stored citations against the current branch to confirm information is still substantiated. By deleting all memories every time a pull request is merged into the main branch.

Which statement best describes the scope of user-level preferences in Copilot Memory?. They are tied to a specific repository and must be recreated when moving to a new project. They are available only to that specific user across all their Copilot interactions in all repositories. They are stored locally on the user's machine and are not available in the Copilot cloud agent. They are available to all users within the same organization to ensure a consistent workflow.

Which Copilot feature only applies stored facts and preferences to the specific user who initiated the operation?. Copilot code review. Copilot CLI. Copilot cloud agent. Copilot for Pull Requests.

Persistent & Non-persistent Copilot SDK. Conversation history. Tool call results. Agent planning state. Session artifacts. Provider/API keys. In-memory tool state.

What is the primary requirement for creating a Copilot SDK session that can be resumed after a restart?. A GitHub Copilot Enterprise plan is required to access the persistence feature. The session must be active for at least thirty minutes before persistence is enabled. The developer must manually provide a unique sessionId during session creation. The developer must call a specific 'saveToDisk()' method after every tool call.

When using Bring Your Own Key (BYOK) with resumed sessions, why must the provider configuration be re-provided?. Resumed sessions always default to the GitHub-managed model provider. The SDK versioning system requires a fresh handshake with the LLM provider upon resumption. API keys are never persisted to disk for security reasons. The provider configuration is tied to the temporary memory of the specific client instance.

In the local filesystem structure for session state, where are the conversation history snapshots stored?. Inside the 'plan.md' file. In the root '~/.copilot/' directory as a single encrypted blob. In the 'files/' directory alongside generated artifacts. In the 'checkpoints/' subdirectory.

How does the Copilot SDK handle concurrent access to the same sessionId by multiple clients?. It uses a filesystem-level lock file to prevent multiple instances from reading the same JSON checkpoints. Concurrent access behavior is undefined, and developers should implement their own locking mechanism. The second client attempting to connect will receive a 'Session Busy' error automatically. The SDK automatically queues requests to the same session in a FIFO (First-In, First-Out) order.

In the context of 'Infinite Sessions,' what does the 'backgroundCompactionThreshold' represent?. The maximum file size in megabytes for the 'session-state' directory. The ratio of context utilization at which the SDK begins compacting the conversation history. The minimum number of tool calls required before history is summarized. The number of days a session remains idle before it is automatically archived.

What happens to a Copilot SDK session that has no activity for 30 minutes?. It is automatically cleaned up (disconnected) by the SDK. The session state is permanently deleted from the disk. The session enters a 'Low Power' mode with reduced reasoning effort. The model is downgraded to a cheaper version to save costs.

Which deployment pattern is recommended for strong isolation and security in multi-tenant environments?. Pattern 2: Shared CLI server. Pattern 3: Global Singleton server. Pattern 4: Peer-to-peer session mesh. Pattern 1: One CLI server per user.

What specifically defines the 'context window' in GitHub Copilot CLI?. The total number of files within the current working directory that Copilot has indexed. A UI element that displays the code currently being edited by the developer. The maximum number of characters allowed in a single user prompt. The total amount of information the AI model can consider at one time when generating a response.

When using the `/context` command, what does the 'Buffer' section represent?. A temporary storage area for messages that were failed to be sent. A reserved portion of the context window that triggers automatic management once reached. The space occupied by the current file content being analyzed. The fixed overhead required for system instructions and tool schemas.

At what percentage of context capacity does the CLI automatically begin the background compaction process?. 100%. 95%. 80%. 50%.

What is the primary outcome of the 'Compaction' process?. Old conversation history is replaced with a structured summary to free up space. Copilot deletes all system instructions to prioritize user messages. The CLI compresses the session data into a `.zip` file in the local workspace. The AI model is swapped for a smaller version that uses fewer tokens.

Which of the following is NOT preserved during a standard compaction process?. The goals of the conversation. The exact wording of every single previous message. Current state of to-do lists. Key technical details and decisions.

What is the function of a 'checkpoint' in Copilot CLI?. It is a validation step where the user must approve the AI's summary. It marks a point in time to which the user can 'undo' code changes. It is a manual save point for the AI's training data. It is a saved copy of a compaction summary that can be reviewed later.

In which scenario would starting a fresh session be more beneficial than continuing a long-running session?. When iterating on a complex bug that requires tracking multiple attempts. When performing exploratory work across a large codebase. When switching to a completely unrelated task. When a multi-phase task requires implementation followed by testing.

What happens if you enter the `/resume` command?. The AI model re-runs the last failed tool command. It forces the AI to ignore the current compaction and go back to raw history. You can pick up a previous session where you left off, including its checkpoints. Copilot generates a summary of the current session without compacting it.

Which Copilot feature is limited to using repository-level facts only, explicitly excluding user-level preferences?. Copilot cloud agent. Copilot code review. Copilot CLI. Copilot Chat in the IDE.