Terraform infrastructure specialist for Vertex AI services and Gemini deployments. Provisions Model Garden, endpoints, vector search, pipelines, and enterprise AI infrastructure. Triggers: "vertex ai terraform", "gemini deployment terraform", "model garden infrastructure", "vertex ai endpoints"
72
60%
Does it follow best practices?
Impact
100%
1.56xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./backups/skills-batch-20251204-000554/plugins/devops/jeremy-vertex-terraform/skills/vertex-infra-expert/SKILL.mdQuality
Discovery
82%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong skill description that clearly identifies its niche at the intersection of Terraform and Google Vertex AI/Gemini services. It lists specific resources and includes good trigger terms. The main weakness is the lack of an explicit 'Use when...' clause, which would improve completeness and make selection criteria clearer.
Suggestions
Add an explicit 'Use when...' clause, e.g., 'Use when the user needs to provision, configure, or manage Vertex AI or Gemini infrastructure using Terraform.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions and resources: 'Provisions Model Garden, endpoints, vector search, pipelines, and enterprise AI infrastructure.' These are specific, named services and components rather than vague abstractions. | 3 / 3 |
Completeness | The 'what' is well-covered (provisions Model Garden, endpoints, vector search, pipelines). However, there is no explicit 'Use when...' clause. The 'Triggers' section partially serves this purpose but doesn't use the standard 'Use when' format, and per the rubric, a missing 'Use when' clause caps completeness at 2. | 2 / 3 |
Trigger Term Quality | Includes natural keywords users would say: 'vertex ai terraform', 'gemini deployment terraform', 'model garden infrastructure', 'vertex ai endpoints'. These cover the key domain-specific terms a user would naturally use when needing this skill, including both the platform (Vertex AI/Gemini) and the tool (Terraform). | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a clear niche at the intersection of Terraform and Vertex AI/Gemini. The combination of infrastructure-as-code tooling with specific GCP AI services makes it very unlikely to conflict with generic Terraform skills or generic AI skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
37%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill provides some concrete-looking HCL examples for Vertex AI infrastructure but suffers from accuracy issues (non-existent or misconfigured resources) and completely lacks workflow guidance for what is inherently a multi-step, potentially destructive provisioning process. It covers only 2 of the 5+ areas mentioned in its description (missing pipelines, Model Garden, enterprise AI infrastructure), and provides no terraform init/plan/apply workflow or validation steps.
Suggestions
Add a clear multi-step workflow with explicit validation: terraform init → terraform plan (review) → terraform apply, including how to verify endpoint health after deployment.
Fix the HCL examples for accuracy: google_vertex_ai_deployed_model is not a standalone resource; deployment is typically done via google_vertex_ai_endpoint_deployed_model or via API/gcloud. Remove the contradictory dedicated_resources + automatic_resources on the same block.
Add coverage for the other promised areas (ML pipelines, Model Garden, enterprise patterns) either inline or via references to separate files.
Include variable definitions and provider configuration so the examples are closer to copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 'What This Skill Does' and 'When This Skill Activates' sections are somewhat redundant with the YAML frontmatter description. The code examples are reasonably lean but the dedicated_resources and automatic_resources blocks being both present on the same deployed model is questionable (they're mutually exclusive in the actual API), adding noise. | 2 / 3 |
Actionability | Provides HCL code blocks that look concrete, but they have issues: google_vertex_ai_deployed_model isn't a real Terraform resource (deployment is done differently), and having both dedicated_resources and automatic_resources on the same resource is contradictory. The vector search example is more plausible but incomplete (missing the index endpoint and deployed index resources needed to actually query it). Missing variable definitions and provider configuration. | 2 / 3 |
Workflow Clarity | There is no workflow sequence at all — no steps for provisioning, no terraform plan/apply commands, no validation checkpoints, no error handling guidance. For infrastructure provisioning (a potentially destructive operation), the absence of any workflow with validation steps is a significant gap. | 1 / 3 |
Progressive Disclosure | Has some structure with headers and a reference link, but the content is fairly thin — only two resource examples with no pointers to additional modules (pipelines, Model Garden, enterprise patterns) that the description promises. The single external reference is appropriate but more navigation to detailed resources would improve discoverability. | 2 / 3 |
Total | 7 / 12 Passed |
Validation
81%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 9 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
allowed_tools_field | 'allowed-tools' contains unusual tool name(s) | Warning |
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 9 / 11 Passed | |
c8a915c
Table of Contents
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.