CtrlK
BlogDocsLog inGet started
Tessl Logo

adk-deployment-specialist

Deploy and orchestrate Vertex AI ADK agents using A2A protocol. Manages AgentCard discovery, task submission, Code Execution Sandbox, and Memory Bank. Use when asked to "deploy ADK agent" or "orchestrate agents". Trigger with phrases like 'deploy', 'infrastructure', or 'CI/CD'.

63

Quality

56%

Does it follow best practices?

Impact

Pending

No eval scenarios have been run

SecuritybySnyk

Advisory

Suggest reviewing before use

Optimize this skill with Tessl

npx tessl skill review --optimize ./plugins/ai-ml/jeremy-adk-orchestrator/skills/adk-deployment-specialist/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Discovery

77%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description does well at specifying concrete capabilities and includes explicit 'Use when' guidance, earning strong marks for completeness and specificity. However, the trigger terms are a weakness: some are overly generic ('deploy', 'infrastructure', 'CI/CD') and would likely cause false matches with general DevOps or deployment skills, while the domain-specific terms (A2A protocol, AgentCard) may not match natural user language.

Suggestions

Narrow the generic trigger terms: replace 'deploy', 'infrastructure', 'CI/CD' with more specific phrases like 'deploy ADK agent', 'Vertex AI agent', 'A2A orchestration', 'multi-agent deployment' to reduce conflict with general DevOps skills.

Add natural user language variations such as 'Vertex AI', 'agent-to-agent', 'multi-agent', 'ADK deployment' to improve trigger term coverage for users who may not use the exact technical terms.

DimensionReasoningScore

Specificity

Lists multiple specific concrete actions: deploy and orchestrate Vertex AI ADK agents, manage AgentCard discovery, task submission, Code Execution Sandbox, and Memory Bank. These are concrete, named capabilities.

3 / 3

Completeness

Clearly answers both 'what' (deploy and orchestrate Vertex AI ADK agents, manage AgentCard discovery, task submission, etc.) and 'when' (explicit 'Use when' clause with trigger phrases like 'deploy ADK agent', 'orchestrate agents', 'deploy', 'infrastructure', 'CI/CD').

3 / 3

Trigger Term Quality

Includes some relevant trigger terms like 'deploy ADK agent', 'orchestrate agents', 'deploy', 'infrastructure', 'CI/CD', but the terms are a mix of domain-specific jargon (A2A protocol, AgentCard) and generic terms ('infrastructure', 'CI/CD') that could overlap with non-agent deployment skills. Missing natural variations users might say like 'agent deployment', 'Vertex AI', 'multi-agent'.

2 / 3

Distinctiveness Conflict Risk

The Vertex AI ADK and A2A protocol specifics create some distinctiveness, but trigger terms like 'deploy', 'infrastructure', and 'CI/CD' are very generic and would likely conflict with general deployment, DevOps, or infrastructure skills. These broad triggers undermine the otherwise specific domain.

2 / 3

Total

10

/

12

Passed

Implementation

35%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

This skill provides a reasonable high-level workflow for deploying ADK agents but lacks the concrete, executable guidance that would make it truly actionable. The instructions read more like a project plan than a technical skill—there are no code examples, no specific CLI commands, and no copy-paste-ready configurations. The progressive disclosure structure is sound in principle but unverifiable without bundle files, and the workflow would benefit from explicit validation checkpoints with error recovery paths.

Suggestions

Add concrete, executable code examples for key steps—especially agent creation (`google.adk.Agent` instantiation), deployment (`vertexai.Client.agent_engines.create()` with actual parameters), and endpoint verification (curl commands or Python snippets to hit `/.well-known/agent-card`).

Include explicit validation checkpoints with feedback loops, e.g., 'Run `adk api_server` locally and verify AgentCard at http://localhost:8000/.well-known/agent-card before deploying' with instructions on what to do if validation fails.

Provide at least a minimal inline example of an AgentCard JSON structure and A2A task submission payload, rather than deferring all examples to a separate file that may not exist.

Add the referenced bundle files (references/errors.md, references/examples.md) or inline their essential content to make the skill self-contained and verifiable.

DimensionReasoningScore

Conciseness

Mostly efficient but the Overview section restates concepts that overlap with the Prerequisites and Instructions. Some phrases like 'Expert in building and deploying production multi-agent systems' are self-descriptive rather than instructive. The content is reasonably tight but could be leaner.

2 / 3

Actionability

The instructions are high-level procedural steps with no concrete code, commands, or executable examples. Steps like 'Implement the agent(s)' and 'Deploy to Vertex AI Agent Engine' are vague directives without specific code snippets, CLI commands, or configuration templates. The Output section mentions `vertexai.Client.agent_engines.create()` but doesn't show actual usage.

1 / 3

Workflow Clarity

Steps are sequenced logically (design → implement → test → deploy → observe), and step 5 mentions validation of endpoints. However, there are no explicit validation checkpoints with feedback loops (e.g., what to do if deployment fails, how to verify each step before proceeding). For a destructive/deployment workflow, the lack of concrete validation gates is a gap.

2 / 3

Progressive Disclosure

References to external files (errors.md, examples.md) and external docs are present and one-level deep, which is good structure. However, no bundle files were provided, so the referenced files (errors.md, examples.md) cannot be verified to exist. The main content itself is thin enough that the references feel like they're hiding essential content that should at least be summarized inline.

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.

Validation9 / 11 Passed

Validation for skill structure

CriteriaDescriptionResult

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

Repository
jeremylongshore/claude-code-plugins-plus-skills
Reviewed

Table of Contents

Is this your skill?

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.