TikTok Ads · Status at source date: Newsletter archive
Google open-sources Agent Skills repository for production agent workflows
Google has open-sourced an Agent Skills repository providing a one-command install of seven production-ready skills into Claude Code, Gemini CLI, Codex, or Antigravity. The skills cover the full agent lifecycle: ADK Python (agents, tools, state management), scaffold (project creation and version upgrades), eval (metrics and LLM-as-judge scoring), deploy (Agent Runtime, Cloud Run, or GKE), publish (Gemini Enterprise registration), observability (Cloud Trace and logging), and workflow (model selection and dev loop).
What changed
Shubham Saboo and Basia Kubicka have both flagged the release.
The structural value is that production-ready agent deployment on Google Cloud: typically weeks of plumbing: now collapses to a single CLI install and a paragraph of intent. For marketing and analytics teams building custom agents on Google's stack, the largest workflow gap has been the friction between prototype and production; this release closes much of it.
Why it matters for advertisers
For you, this means that marketing-ops teams considering custom agent deployment on Google Cloud should evaluate open-sourced skills as a starting point rather than building scaffolding from scratch. Particularly relevant for measurement and reporting agents where the deployment, observability, and evaluation layers are the parts most teams skip when prototyping.
Perspective from the original PMC newsletter.Sources & contributor credit
- Workflow explanation · Checked Basia Kubicka, workflow and image source
Inspected for attribution on 11 September 2026. This date is the review date.
- Newsletter coverage · Paid Media Collective newsletter
Original newsletter text, contributor labels and media for this update.
- Source referenced in newsletterLinkedIn
Linked from the original newsletter. The source publication date has not been independently confirmed.
Attribution needs clarification
Basia Kubicka supplied the linked seven-skill workflow explanation and credited image source. Shubham Saboo discusses a separate Google skills repository. Their coverage should not be treated as authorship of the underlying code.
Attribution evidence and limitations
Read both original posts. Basia describes agents-cli and seven enterprise-agent skills; Shubham describes google/skills and a different product-skill list. The original newsletter combines these descriptions.
The newsletter explicitly labels Basia’s link as its image credit. Both profiles are linked from their respective posts. Code authorship is unresolved in this record.
Visually inspected the graphic, which shows the agents-cli README and architecture. It has no individual design credit; Basia is retained as the credited image source, not asserted as the code or diagram’s original creator.
Attribution checked . This is a review date, not the original publication date.
- Published on this site
This update reflects the dated source reporting. Availability may have changed. Further coverage of this same development will be added to this page.
Original newsletter text and archive evidence
Google open-sources Agent Skills repository for production agent workflows
Google has open-sourced an Agent Skills repository providing a one-command install of seven production-ready skills into Claude Code, Gemini CLI, Codex, or Antigravity. The skills cover the full agent lifecycle: ADK Python (agents, tools, state management), scaffold (project creation and version upgrades), eval (metrics and LLM-as-judge scoring), deploy (Agent Runtime, Cloud Run, or GKE), publish (Gemini Enterprise registration), observability (Cloud Trace and logging), and workflow (model selection and dev loop). Shubham Saboo and Basia Kubicka have both flagged the release.
The structural value is that production-ready agent deployment on Google Cloud: typically weeks of plumbing: now collapses to a single CLI install and a paragraph of intent. For marketing and analytics teams building custom agents on Google's stack, the largest workflow gap has been the friction between prototype and production; this release closes much of it.
For you, this means that marketing-ops teams considering custom agent deployment on Google Cloud should evaluate open-sourced skills as a starting point rather than building scaffolding from scratch. Particularly relevant for measurement and reporting agents where the deployment, observability, and evaluation layers are the parts most teams skip when prototyping.
Source captured . No explicit first-contributor label was provided for this update.
