BB Google Cloud
Architects, debugs, secures, and cost-optimizes Google Cloud — Cloud Run, GKE, Compute Engine, BigQuery, Cloud SQL, IAM, VPC. Use when deploying or reviewing anything on GCP, when a bill jumps or BigQuery scan cost has to come down, when a 403 PERMISSION_DENIED, a SERVICE_DISABLED, a RESOURCE_EXHAUSTED quota error, a 502/503/504, or an unreachable Cloud SQL instance has no obvious cause, when choosing between compute (Cloud Run, GKE, Cloud Functions) or databases (Cloud SQL, AlloyDB, Spanner, Firestore, Bigtable), when hardening service accounts, org policies, firewall rules, public buckets, or secrets, when writing Terraform against the google provider, or when auditing an inherited project. Covers VPC/subnet design, Private Google Access, Vertex AI/GPU quota, Pub/Sub/Dataflow, backups/DR, and gcloud. Not for Kubernetes manifest authoring (`k8s`), Terraform language mechanics (`terraform`), or PostgreSQL tuning inside Cloud SQL (`pg`).
As a process B 68/100 · Nearly there — weak spots: inputs and preconditions, consistency, execution cost
How to improve
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 1
✓ No critical or high findings
Medium and low: 1
-
low Risky intent
intent-offensive-securityiam.md:42Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (detector / deny-list definition)This dual nature is the source of most GCP IAM confusion, and of most GCP privilege escalation.
detector
Files scanned: 21. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
name-formatname should be kebab-case (lowercase letters, digits, hyphens) - warning
body-longSKILL.md body ≈ 8801 tokens (recommended < 5000); move details to references/ - note
frontmatter-keyunknown frontmatter key "slug" - note
frontmatter-keyunknown frontmatter key "homepage" - note
frontmatter-keyunknown frontmatter key "changelog"
Process rating: all ten parameters 68/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 40Consistency. Frontmatter name (Google Cloud) differs from the folder (gcp)
- 40Execution cost. Instruction body is 8801 tokens: crowds the task out of the window
- 60Tools and files. Uses tools (bash, node) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 100Steps. 41 steps
- 100When it triggers. States when to use and when not to
- 100Failures and branches. 6 branches, has a failure section
- 100Running it twice. Mutating operations check current state
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 15 top-level sections: this looks like several domains in one skill
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +3Description length 950: 120–800 characters recommended
- +4No input/output examples
- +1No license
- +2Single-language instructions
- +4Description says when NOT to use the skill
- +4Structure: 15 headings
- +3Step-by-step instructions: 41 items
- +3Output format is stated explicitly
Quality base 70; lint remarks subtract, signals add up to 100. Result: 66.