Key takeaways

  • Google announced Gemini skills on September 30, 2026 and is gradually rolling them into Gemini chat. The launch note says availability spans Google AI subscription tiers for adults; the live help page currently limits creation and use to personal accounts with Keep Activity enabled.
  • A skill stores reusable instructions and can be invoked with “/”, selected automatically from prompt context or combined with other skills. Google also supports importing a folder containing SKILL.md plus optional reference files.
  • Google plans to remove Gems in stages: November 2026 for personal accounts, March 2027 for Workspace business, enterprise and nonprofit accounts, and June 2027 for education accounts. Existing Gems are supposed to be recreated automatically as skills.
  • The transition is not feature-complete today. Workspace support, sharing, Drive files and Gemini Notebook integration are promised for the coming weeks, while Google lists current editing defects on mobile and Mac.
  • Inventory and export important Gems before the applicable deadline, then validate migrated skills with explicit test prompts, file checks and automatic-activation tests rather than assuming one-to-one behavior.
01

Google announced the Gemini chat rollout on September 30

Google announced on September 30, 2026 that skills are rolling out directly in Gemini chat globally. The launch footnote says they are available across Google AI subscription tiers for people 18 and older, with under-18 access planned later. Google says Workspace business, enterprise, nonprofit and education availability will follow in the coming weeks; the current help page still says work and school accounts cannot use the feature.

This is a Gemini product and workflow update, not a new model or API release. Google does not name a model ID, publish a developer endpoint, quote a separate skills price or provide service-level limits in the reviewed pages. Account eligibility and rollout state should be checked in the actual Gemini client rather than inferred from the global announcement.

02

Skills turn repeated prompts into reusable workflow components

A skill stores custom instructions that can be reused across chats and tasks. A user can type “/” and choose a skill, while Gemini can also activate a relevant enabled skill automatically from the context and intent of a prompt. Multiple skills can be combined in one task, and one skill can reference another.

Google says skills can be built from a chat, with Gemini’s help, from a template or by uploading a file or folder containing SKILL.md and optional supported reference files. The launch announcement names text documents, PDFs and images as reference inputs. These are useful packaging features, but they do not establish deterministic execution, output accuracy or a stable developer contract.

03

Current access has account, age, activity and client boundaries

The live help page says users must be 18 or older, sign in with a personal Google Account and keep Gemini Apps Activity enabled. It describes a gradual rollout, so an eligible account may not see skills immediately. The listed clients are the Gemini web app, Gemini mobile app and Gemini app on Mac.

Google currently flags two defects: manually created skills do not save in the mobile app, and skills with uploaded files cannot be edited in the mobile or Mac apps. Its temporary path is to create conversationally or use the web app, and to edit file-backed skills on the web. Teams should treat client parity as unverified until these notices are removed and their own account passes a smoke test.

04

Gems now have three retirement windows

Google says Gems support will start ending in November 2026 for personal accounts, March 2027 for Workspace business, enterprise and nonprofit customers, and June 2027 for Workspace education customers. It plans to recreate existing Gems as skills automatically when Gems go away. Google also says sharing, Google Drive files and notebooks from Gemini Notebook are coming to skills in the weeks around the transition.

Automatic recreation is a migration mechanism, not evidence of behavioral equivalence. The separate launch footnote says experimental “Gems by Google Labs” will not migrate. Google also says Opal will turn down with the personal-account change in November. Owners should therefore identify the exact product and account type behind each workflow before relying on automatic migration.

05

Skills in chat are not the same contract as Gemini Spark

Skills were already available in Gemini Spark before this chat rollout. Spark is Google’s separate personal agent surface for managing tasks and schedules using connected apps, websites, chats, location and other sources. Its help page requires a personal account, Keep Activity and a Google AI Pro or Ultra subscription, and currently excludes the European Economic Area, Nigeria, Switzerland and the United Kingdom.

Do not transfer Spark’s price or regional rules wholesale to ordinary Gemini chat skills. The reviewed chat-skills help page does not list the same subscription or regional prerequisites, while the launch footnote references Google AI subscription tiers. The documentation therefore supports reporting distinct surfaces, not claiming universal free access or identical availability.

06

Automatic activation raises a control question

Convenient automatic selection also means an instruction bundle can affect a response without the user naming it in that prompt. Before enabling a skill automatically, test prompts that should activate it, near-miss prompts that should not and conflicts between stacked skills. Record which skill ran, which files it could use and what output changed.

Reference files and cross-skill composition expand the context boundary. Keep confidential material out unless the account’s data controls and intended users are approved. Separate style and formatting instructions from any workflow that could send messages, modify files or take other consequential actions, and require an explicit review or approval step where an error carries material cost.

07

What Gem owners should do before migration

Inventory every Gem by owner, account class, instructions, knowledge files, sharing scope and business dependency. Export or separately retain critical instructions and source files before the applicable retirement window. Recreate high-value workflows early in a noncritical setting, pin a small acceptance set and compare outputs from the Gem and skill without silently changing the evidence or model configuration.

Adopt skills now for low-risk reusable instructions when the account has access and the web client passes a save-and-reopen check. Constrain automatic activation until trigger and conflict tests pass. Wait for promised sharing, Drive, Notebook or Workspace support when those are requirements. Replace the workflow rather than migrate it unchanged if it depends on undocumented behavior, unavailable files or unattended consequential actions.

08

Copy-ready Gem-to-skill migration record

Complete one record for each Gem or reusable instruction set before its account-specific retirement window.

Entries stay in this browser tab and are not submitted to AccessAllGPT. Blank responses are copied as [Unresolved].

Gem name, owner, account class, users, purpose, business consequence and applicable November, March or June deadline.

Exported instructions, knowledge files, sharing list, connected data, example prompts, expected outputs and last verified date.

Creation path, SKILL.md or UI instructions, reference files, linked skills, client used and save-and-reopen evidence.

Explicit only or automatic; positive triggers, near misses, exclusions, conflicts, disable path and user-visible indication.

Representative prompts, Gem baseline, required content, prohibited behavior, file citations, repetitions and reviewer.

Eligibility, Keep Activity, subscription, region, retention, sensitive files, access, sharing and administrator approval.

Web, mobile and Mac behavior; file editing; known-issue status; Workspace availability and promised-feature dependencies.

Connected apps, write actions, schedules, approvals, audit evidence, failure behavior and rollback.

Adopt, constrain, wait or replace; approved scope, owner, migration date, unresolved gaps and next review.

Primary sources

  1. Let skills in Gemini tackle your most repetitive tasksGoogle · Reviewed: September 30, 2026 publication date; Gemini chat rollout; invocation and automatic-use behavior; skill stacking; reference files; Workspace timing; Gems retirement and migration dates; subscription and age footnotes · Retrieved · Supports: Google announced a global Gemini chat rollout for reusable skills, said the launch covers Google AI subscription tiers for adults, and published staged Gems retirement dates beginning with personal accounts in November 2026.
  2. Create & manage skills for Gemini AppsGoogle Gemini Apps Help · Reviewed: Current rollout notice; eligibility and activity prerequisites; supported clients; SKILL.md and reference-file imports; automatic and explicit invocation; skill composition; known mobile and Mac editing issues · Retrieved · Supports: Google documents skills for adults using personal Google Accounts with Keep Activity enabled, gradual availability across Gemini web, mobile and Mac, explicit or automatic invocation, composition, SKILL.md imports and two current client limitations.
  3. About the transition from Gems to skillsGoogle Gemini Apps Help · Reviewed: Transition overview; personal, business, nonprofit and education timelines; automatic Gem recreation; manual recreation workflow; knowledge-file handling; skills capabilities and transition FAQs · Retrieved · Supports: Google says existing Gems will be automatically recreated as skills when Gems are removed, while also documenting a manual migration path and separate retirement windows for personal and organizational accounts.
  4. Use Gemini Spark to manage your tasks & workflows in Gemini AppsGoogle Gemini Apps Help · Reviewed: Spark definition; schedules and tasks; personal-account and subscription prerequisites; regional exclusions; supported clients; skill use; connected information sources; safety notices · Retrieved · Supports: Google documents Spark as a separate personal AI agent for tasks and schedules that can use skills, requires a Google AI Pro or Ultra subscription, and currently carries regional and account restrictions distinct from skills in ordinary Gemini chat.

Limitations

AccessAllGPT reviewed four public first-party Google pages but did not receive the Gemini skills rollout, create or run a skill, inspect SKILL.md parsing, upload reference files, migrate a Gem, test automatic activation or skill composition, verify model selection, compare outputs, exercise Workspace administration, use Gemini Spark, confirm account-specific subscription access, inspect data handling or reproduce current client defects. Google describes the rollout as gradual and promises several capabilities for coming weeks; those are not treated as currently universal. The reviewed sources do not publish a skills API, model ID, separate price, rate limit, service-level commitment or complete versioned execution contract. Product behavior, availability and retirement details can change.

Disclosures

AccessAllGPT received no Google account access, subscription, credits, briefing, demo, migration support, review or compensation for this article. Google did not sponsor, review or endorse it. AccessAllGPT Research is operated by NeuralArc, is independent, and is not affiliated with Google. Publication-wide relationships are listed on the disclosures page.

Further AccessAllGPT guidance

  1. Google Launches Guided Vision in Gemini Live on Android
  2. Google Announces Gemini 4 Argon, but Broad Access Is Still Pending
  3. OpenAI Launches Dots, Always-On GPT-6 Astra Agents
  4. Human Approval Boundaries for AI Agents
  5. Move an LLM API Without Breaking Production
  6. AI API Data Retention and Residency Procurement Gates
  7. AccessAllGPT Research methodology
  8. Publication disclosures

Continue the research

Get evidence-led updates for teams making production AI decisions.