What does this skill do, and when should you use it?
This skill provides procedures for creating, listing, describing, updating, and deleting Google Agent Platform serving endpoints. It treats endpoints as logical serving hosts that provide stable URLs for online predictions. It requires Google Cloud authentication, an active project, and an explicit region for commands. It does not deploy models or run model evaluations.
Uses the gcloud CLI to authenticate with Google Cloud, configure application default credentials, and set the active project; lists and describes endpoints in a specified region; creates or updates endpoints after the required confirmation; deletes endpoints only after explicit typed confirmation; directs traffic-split work to the model deployment flow; and gives troubleshooting guidance for permission, quota, and resource-busy errors.
- A Google Cloud user needs to discover endpoint IDs in a particular region.
- A platform engineer needs to create an endpoint before deploying a model for online prediction.
- An administrator needs to change an endpoint display name or labels with a confirmation step before mutation.
- An operator needs to diagnose a failed endpoint deletion, permission error, or regional quota problem.
How do you install this skill?
- Deletion is irreversible, and no recovery or backup procedure is documented; endpoint preflight checks are intentionally deferred until after confirmation.
- The authentication steps invoke gcloud login and ADC setup; users should verify credential provenance, IAM scope, and project scope before use.
- The traffic-splitting section is incomplete and depends on another skill outside this assessment scope.
- Chinese-language support and mainland-China network reachability are undocumented.
- Shell / CLI
- Network access
gcloud CLI
Run npx skills add google/skills and select Agent Platform Endpoint Management during installation. The skill is located at skills/cloud/agent-platform-endpoint-management/SKILL.md.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/google/skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/cloud/agent-platform-endpoint-management" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- List Agent Platform endpoints in us-central1
- Describe endpoint ENDPOINT_ID.
First run gcloud auth login and gcloud auth application-default login, then gcloud config set project $PROJECT_ID. Provide a region in the request, such as “List Agent Platform endpoints in us-central1” or “Describe endpoint ENDPOINT_ID.” Listing and describing require no confirmation; creation and updates require showing the complete literal command and waiting for Yes/No; deletion requires immediate typed confirmation before any pre-flight checks.
What are this skill's strengths and limitations?
- Covers endpoint listing, inspection, creation, updating, and deletion.
- Defines separate confirmation rules for read-only, reversible, and destructive actions.
- Provides concrete guidance for permission, regional quota, and resource-busy failures.
- Requires the gcloud CLI, Google Cloud authentication, an active project, and an explicit region.
- Does not handle model deployment, undeployment, or model evaluation.
- The supplied source gives no test suite, version compatibility matrix, or automation scripts.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Agent Platform Endpoint Management this page ✓ Google · Official | 50 · Use with care | ★ 21k | 12d ago | Apache-2.0 |
| Agent Platform Model Deployment Assistant ✓ Google · Official | 43 · Not recommended | ★ 21k | 12d ago | Apache-2.0 |
| Google Cloud gcloud CLI Safety Skill ✓ Google · Official | 51 · Use with care | ★ 21k | 12d ago | Apache-2.0 |
| GKE Basics Navigator ✓ Google · Official | 45 · Not recommended | ★ 21k | 12d ago | Apache-2.0 |
| Google Cloud Global Load Balancer Configurator ✓ Google · Official | 45 · Not recommended | ★ 21k | 12d ago | Apache-2.0 |
How did FollowSkills review this skill?
The document clearly separates read-only, reversible mutations, and deletion, requiring confirmation for create, update, and delete, with deletion confirmation before preflight checks. However, the authentication commands are broad, least-privilege IAM guidance and credential/data-flow handling are absent, and there is no deletion recovery or audit-verification procedure, so points are deducted.
The main list, describe, create, update, and delete command paths are mostly coherent and include region requirements plus some error categories. However, the traffic-splitting section provides no actual update command and delegates to another skill; authentication, invalid parameters, resource states, and post-failure diagnostics are thin, so the static score is limited.
The name, purpose, CRUD operations, and exclusions for model deployment and evaluation are reasonably clear, and regional input is required. Trigger precision, non-fit boundaries, Chinese-language support, and mainland-China network reachability are not addressed, while traffic-splitting coverage is incomplete, so points are deducted.
The skill includes metadata, overview, safety tiers, operation sections, troubleshooting, and examples; repository material supplies Apache-2.0 licensing, installation, support channels, and an active-development note. It lacks skill-level versioning, changelog, named maintenance ownership, dependency notes, stable parameter contracts, and fuller limitations/FAQs, so it does not receive full marks.
It provides directly usable gcloud examples for regional endpoint listing, inspection, creation, updating, and deletion, covering the main management task. But representative outputs, rollback, complete traffic-splitting instructions, and sufficient result validation are missing, leaving users to perform additional judgment and troubleshooting, so points are deducted.
The skill and repository text provide auditable commands, scope, and governance information, but contain no key-path test suite, CI execution evidence, independent corroboration, or reproducible examples. The assessment therefore rests primarily on static text and receives only limited credit.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →