Gemini skills add files and sharing as Gems retire | Magica
Gemini skills add files and sharing as Gem migration leaves gaps
Editorial Team
••📖6 min read
Google is rolling out file references and sharing for Gemini skills as adult personal accounts lose the ability to use Gems. Automatic conversion preserves supported material, but file limits and missing tools can interrupt existing workflows.
Google is rolling out file references and sharing for Gemini skills, its reusable custom instructions.
Gems became read-only for personal-account users aged 18 and over on November 17; access ends December 17, 2026.
Conversion can leave out files, and skills still lack several tools available in Gems.
Google is adding live file references and sharing to skills in its Gemini AI assistant as it retires Gems, the custom assistants those skills replace. The additions restore capabilities familiar to Gem users, but automatic migration does not guarantee that an existing workflow will survive intact.
The immediate stakes are practical: affected personal-account users can no longer run their Gems, while some attachments and built-in tools are unsupported in skills. They have until December 17 to recover files that fail to transfer.
Google’s schedule separates the adult personal-account read-only date from removal, with later transitions for Workspace accounts. Graphic: Magica; source: Google Gemini Apps Help. Source: Google Support.
A new way to use saved instructions
Gems already let users save instructions for recurring work. Google launched them in August 2024 for Gemini Advanced, Business and Enterprise subscribers as custom versions of Gemini that users could name, instruct and return to for specialized help.
Skills change how those instructions enter a conversation. Users can invoke them within an existing chat, combine several, or let Gemini automatically apply an active skill when relevant. Google cites that flexibility in its explanation for replacing Gems. Reusable instructions are the continuity; applying several within the same conversation is the change.
The latest additions let skills reference files from a device, Google Drive, Gemini Notebook and Google Photos. Google says Gemini consults those files whenever the skill runs in chats and tasks, and that Drive files and notebooks update in real time, as described in the November 16 report on the rollout.
Microsoft is pitching Copilot Workspaces as a shared environment for people and AI agents. Team collaboration is already in early access, but personal Autopilot integration is still coming soon, with final agent pricing unannounced.
Editorial Team
Google’s November announcement says the gradual rollout for Workspace release domains began November 12, with up to 15 days for feature visibility. It lists availability for all customers of Workspace, Google’s work and school productivity service, and users with personal Google accounts. Availability across those account groups does not mean everyone receives the features at once.
Sharing instructions does not grant file access
Owners can share a skill from Customize → Skills in the Gemini web app, granting Viewer or Editor access. Viewers can use it and inspect its description, instructions and uploaded files. Editors can change instructions and uploaded files and share the skill with others, according to Google’s skills documentation.
Sharing and access management are currently web-only. Notification emails are optional; a newly shared skill appears as inactive for its recipient and must be reviewed and added before use. Sharing the skill also requires separately sharing access to its reference files.
Workspace administrators retain another layer of control. Google says skill sharing is enabled by default but can be disabled in the Admin console, and follows the organization’s Google Drive sharing settings. External sharing therefore depends on whether the organization permits Drive files to be shared outside it.
Connected workflows have a further condition: Google’s documentation says skills work with Keep Activity off, but features requiring that setting, including Connected Apps, do not. Skills cannot access Workspace services such as Drive and Calendar while it is off.
Portability also has limits. Google allows imports from other platforms through a SKILL.md instruction file, but scripts cannot make requests to external websites. Downloading a skill excludes its connected Drive files and notebooks. A saved copy of a shared skill does not automatically sync with later changes to the original. Moving the instructions is therefore different from preserving all their dependencies.
Skills have a 100 MB combined file limit; Google says conversion removes the largest files first until under that limit. Graphic: Magica; source: Google Gemini Apps Help. Source: Google Support.
Automatic conversion can leave material behind
The November 17 transition report describes Gems becoming draft skills with instructions, descriptions, sharing permissions and supported knowledge files. Google’s transition guidance confirms automatic recreation, supported-file transfer and preservation of sharing permissions. Gems previously shared with a user appear as inactive skills for review and acceptance.
The file qualification matters. Skills have a 100 MB combined file limit. Google says that when a Gem exceeds it during conversion, the largest files are removed first until the result is under the limit. That is a total per skill, rather than an allowance for each attachment.
Google also lists these Gem attachments as excluded from conversion:
GitHub files, audio and video.
Legacy Office formats: .doc, .xls and .ppt.
Compressed archives, including .zip, .tar and .rar.
Modern Office formats, PDFs, text and images are among the supported files, although the skills documentation lists additional exclusions for some code and data formats. Users need to check the actual file list rather than assume every document or script transfers.
Tool support is a separate gap. Google says most default Gem tools do not yet work with skills, including Canvas, Deep Research, Guided Learning, and video and music creation. Connected Apps can work with skills, but do not establish parity with those built-in tools. A workflow based on saved writing instructions may transfer more readily than one requiring an unsupported research tool or attachment.
What needs checking before access ends
Google’s schedule makes November 17 the read-only transition date for personal-account users aged 18 and over: Gems remain accessible, but cannot be used or edited, and new ones cannot be created. On December 17, 2026, those Gems and their files will no longer be available in Gemini Apps.
Google advises opening the old Gem and new skill side by side, identifying files that did not transfer and downloading them before that deadline. Shared permissions carrying over is not the same as recipients having accepted and activated the new skill.
Other groups have more time. Google lists March 2027 for Workspace business, enterprise and nonprofit accounts, and June 2027 for education accounts. Personal-account users under 18 will transition later, on an unspecified date.
The unresolved question is when skills will support workflows that depend on the missing tools. Google gives no restoration dates for them in its transition guidance. Until it does, the meaningful test is whether a converted skill retains the needed files, access and capabilities—not simply whether it appears on the Skills page.
Muddy Waters disclosed a short position in Nvidia while questioning a possible connection involving senior adviser Ajay K. Puri. Its own account leaves the key identity question unanswered, separating its new claims from an existing server-diversion prosecution.
Nvidia reported $115.8 billion in quarterly revenue in its first earnings release since co-founder Jensen Huang resigned. Strong demand gives interim CEO Colette Kress a financial cushion, while a campaign for Huang’s return collides with unresolved investigations and the board’s reported opposition.
OpenAI launched GPT-6.2 Astra and Sol for professional work, cutting Astra’s standard input and output token prices by 20%. The models offer cheaper paths through selected tasks, but research evaluations, reporting failures and unfinished independent safeguard testing complicate the case for broader autonomy.
Microsoft’s first live speech-to-speech model offers OpenAI API compatibility and planned enterprise integrations. Its published evaluations show a low-cost option, with several rivals completing more agentic tasks.
Canberra-based Trellis Data is offering its 220-billion-parameter Agentaus 1 model to business and government. Its pitch goes beyond keeping data in Australia, but performance, deployment costs and a planned capital raising remain key tests.
Perplexity is adding direct collaboration between frontier models to Computer, letting a lead agent coordinate up to 16 supporting agents. Company-reported benchmarks suggest mixed teams can outperform repeated copies of one model, but leave customer costs and production reliability unresolved.
Marsh reported 105 client cyber incidents in the last two weeks of October, against 48 in the final two weeks of May. The broker links the increase partly to AI, while warning that insurance panel access may not guarantee an available response team.
Meta’s Muse personal AI agent has launched in Australia, with some local shopping and booking connections to follow. Its browser and payment tools broaden its reach, while the company’s technical disclosures qualify its privacy promises.
Microsoft and Salesforce are extending their sales software alliance with Copilotforce, combining CRM workflows with Microsoft 365 context under existing permissions. The Copilot plugin enters public preview, while Teams remains in pilot; earlier Microsoft tools and a Claude beta raise the bar for proving its value.
Microsoft expands MAI Image 2.7 from an editing preview into image generation and API access, with cheaper Flash pricing and a Canvas workspace. Benchmark claims depend on the workload, while the MAI Labs interface carries restrictions on commercial use.
OpenAI is widening applications for its legal AI configuration after improving research scores and expanding its search index. Access still requires approval, and benchmark results show substantial limits alongside the gains.
Microsoft has made Word, Excel and PowerPoint creation generally available inside Copilot, combining editable files with workplace permissions and collaboration. The broader shift toward autonomous work remains staggered across previews, while its claimed cost advantage rests on company-funded testing.
Knit Health’s NEA-led Series A funds deployment of Med-BEAM for care navigation and hospital operations. Its early company benchmark beats three Claude models at predicting clinical actions, while better patient outcomes remain unproven by that test.
Science Corporation’s Series D follows its European retinal-implant launch. Clinical results support improved central vision, but U.S. approval, eligible patients and commercial terms remain consequential questions for the startup, reportedly valued at $3.5 billion.
Google will let Gemini users save conversations while opting out of generative AI training on future chats. The global rollout also extends storage of history-off and temporary chats from 72 hours to around two months; human review and longer retention can still apply.
Google is removing the subscription requirement for Gemini’s task mode in six countries, while adding U.S. payment, calling and messaging features. The expansion broadens access to delegated work, but usage limits, permissions and website compatibility still govern what it can finish.
Eligible Pixel owners can join a Play Store beta to generate home-screen widgets from descriptions, but device, language and regional restrictions still apply.
Microsoft’s new in-house coding models compete with outside suppliers inside GitHub Copilot. Company tests show lower task costs at promotional prices, while its continuing financial ties to OpenAI complicate the rivalry.
Valkryn disclosed a $15 million seed round and early military testing contracts. Its evaluator-focused Battle Lab faces existing commercial tools, public testing capabilities and funding constraints that software alone cannot resolve.