GitHub Copilot Memory Is GA: Check and Delete Context Before Enabling Dreaming
Memory carries preferences across sessions, while Dreaming can update them, consume AI credits, and pair with cross-repository search
Key takeaways
GitHub announced GitHub intelligence on October 28, 2026. Copilot Memory is generally available on paid plans, Dreaming is an opt-in public preview, and code search can reach related repositories and dependencies. Dreaming may add, update, or remove personal and repository memories, while proposed changes to checked-in skills and instructions arrive as pull requests for team review. Dream runs consume AI credits. This article gives developers a practical way to inspect and delete saved preferences, separate personal from repository context, limit repository access, and decide whether Dreaming and cross-repository search belong in a low-risk trial before broader adoption. The trial remains reversible.
By ENHE AI · 2026-11-02 · AI-assisted drafting and editorial checks.
Direct answer: Treat memory as reviewable configuration before letting it evolve
GitHub announced GitHub intelligence on October 28, 2026, and Copilot Memory is generally available on paid Copilot plans. It can carry personal preferences and repository conventions across sessions. That reduces repeated prompting, but it can also repeat a stale preference, so the first step is to inspect and clean existing memories.
What Memory, Dreaming, and code search each do
Memory stores personal preferences and repository conventions. Copilot Dreaming is in public preview. Dreaming requires opt-in by a user or administrator. It curates context from past work, while code search can find text, symbols, patterns, and related code across repositories and dependencies without a local clone.
A reversible enablement workflow
Start in Copilot settings by viewing or deleting inaccurate personal memories. Choose one repository without production secrets and record its current skills and instructions. Enable Dreaming only there, complete one low-risk task, and compare memory changes, proposed pull requests, and AI-credit use. Disable it and delete new memories if the result is not useful.
Permission, cost, and human-review boundaries
Dreaming can add, update, and remove personal and repository memories. For checked-in skills and instructions, it proposes pull requests for team review. Dream runs consume AI credits. Enterprises should also limit repository access so cross-repository search does not bring irrelevant or sensitive context into the current task.
Frequently asked questions
Does Memory GA mean Dreaming is on by default?
No. Memory is generally available, while Dreaming remains an opt-in public preview with separate controls for individuals and enterprises.
Can an incorrect preference be deleted?
Yes. Personal memories can be viewed or deleted in Copilot settings, and repository-level changes should also be checked through their pull requests and settings.
Why this matters
AI coding is moving from one-off prompts toward persistent context. The benefit is less repetition; the risks are stale memories, broader repository scope, and the cost of automated curation. Treating memory as configuration makes the workflow more controllable.
Related tools and reading
These pages support the same task, workflow, or risk boundary.
- Repository plugin defaults and enterprise controls
- Copilot cloud sessions and shared entitlement
- A safe first run for Agentic Workflows
Conclusion
Clean up personal memories first, then enable Dreaming in one non-sensitive repository. Record what it adds, changes, or removes, and check AI credits against official billing guidance and information actually available in the account. If no usage data is visible, record only that the run is known to consume credits. Expand only when memories are traceable, repository scope is explicit, and pull-request review is effective.
What this means for everyday users
Dreaming can add, update, and remove personal and repository memories. For checked-in skills and instructions, it proposes pull requests for team review. Dream runs consume AI credits. Enterprises should also limit repository access so cross-repository search does not bring irrelevant or sensitive context into the current task.
Related tutorials
[[zh]][[/zh]][[en]]Review a public website[[/en]]
SearchScope SEO/GEO
[[zh]][[/zh]][[en]]Access and usage guide[[/en]]
AI Prompt Management System | 418 Bilingual Prompts for Writing, SEO and AI Creation
High-Frequency AI Prompts for Work, Learning, and Teaching guide
High-Frequency AI Prompts for Work, Learning, and Teaching
Related reading
GitHub Agentic Workflows Is GA: Start With a Read-Only Issue Triage Task
GitHub made Agentic Workflows generally available on October 28, 2026. Teams can describe issue triage, CI failure analysis, documentation updates, or test writing in natural-language Markdown, then compile the definition into standard GitHub Actions YAML that reuses existing runners and policy constraints. Workflows use read-only permissions by default, execute in a sandboxed container behind a network firewall, and apply safe-output rules and threat detection. The gh aw forecast command now includes threat-detection credits and monthly AI-credit ranges. This tutorial starts with a read-only issue-triage workflow and explains the entry conditions, inputs, expected output, budget evidence, review gate, and rollback path without claiming unobserved hands-on results.
GitHub Copilot Cloud Sessions Are GA: How Cross-Device Work Uses Actions Entitlement
GitHub made its connected Copilot cloud experience and cloud sandboxes generally available on paid Copilot plans on October 28, 2026. A user can start a task on github.com, GitHub Mobile, Copilot CLI, the Copilot app, VS Code, Slack, or Microsoft Teams, then continue the same session elsewhere. The included Actions compute entitlement becomes one shared, dollar-based pool for standard hosted runners, cloud sandboxes, and certain larger runners. Billing for cloud-sandbox usage begins in November 2026, so cross-device convenience can affect the same budget used by builds. This article helps users check repository scope, write access, tool permissions, entitlement consumption, and stop conditions before their first cloud session.
GitHub Copilot Repository Plugin Defaults Are GA: Prevent Permission Drift Across Teams
GitHub made repository plugin defaults and enterprise-managed plugin controls generally available in VS Code, Copilot CLI, and the GitHub Copilot app on October 28, 2026. A maintainer can define a marketplace and default plugins for a trusted repository so contributors avoid repeating setup. Enterprise administrators can require or block plugins and restrict installation to approved marketplaces. Managed settings take precedence over repository and personal choices, and a required plugin cannot be disabled by the user. This article gives teams a rollout checklist for validating publisher identity, plugin contents, MCP connections, credentials, uninstall behavior, repository trust, and the difference between a convenient default and a mandatory enterprise control.
From Chat Boxes to Personal AI Companions: AI Assistants Are Entering the Desktop Execution Era
AI assistants are moving from answering questions toward continuing real tasks. AI agents, MCP tool ecosystems, personal memory, and local workbenches are pushing this shift together. For users, the real value is not another chat box, but less repeated context setup and more continuity from thinking to doing.
VS Code Agent Host Runs in the Background: Choose Phone Control, Local Machines, or Cloud Sessions
GitHub announced its VS Code Copilot Universe release on October 28, 2026. The new Agent Host lets sessions continue after a project folder or editor window closes, can run on another machine through Remote SSH or dev tunnels, and can be directed from GitHub Mobile. A local or remote machine and its Agent Host must remain running when the work executes there. Computer use is experimental and requires approval for each application. The open Agent Host Protocol lets custom clients reconnect to the same live session. This article compares local, remote, and cloud placement, then provides a first-run checklist for availability, permissions, desktop access, and recovery.
GitHub Issue Artifacts Preview: What Agent Outputs Can Store and How Version History Works
GitHub announced Issue Artifacts in public preview on October 28, 2026. Teams can attach agent-generated plans, research, links, visualizations, and interactive HTML directly to an issue. Updates made on github.com retain version history as an artifact changes. GitHub CLI adds gh issue artifact commands to create, edit, list, view, download, and delete artifacts, while Markdown artifacts can include image and video attachments. This explainer uses a root-cause-analysis example to show which outputs belong beside an issue, how to separate observed evidence from unverified conclusions, when history helps reviewers follow a decision, and when an incorrect or sensitive artifact should be corrected or removed rather than being treated as repository truth.