Name: benchmark-agents
Rating: 56.00000000000001 (1 reviews)
Author: vercel

benchmark-agents

Advanced AI agent benchmark scenarios that push Vercel's cutting-edge platform features — Workflow SDK, AI Gateway, MCP, Chat SDK, Queues, Flags, Sandbox, and multi-agent orchestration. Designed to stress-test skill injection for complex, multi-system builds.

Quality

62%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./.claude/skills/benchmark-agents/SKILL.md

Quality

Content

78%Weight 40%Scale 1-5

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

The body is highly actionable and well-sequenced, with concrete commands and explicit verification/release gates that make the eval loop easy to execute. Its main weaknesses are minor redundancy in the prohibition rationale and a monolithic single-file structure with no progressive disclosure for the large reference tables.

Suggestions

Move the scenario table and common-issues table into one-level-deep reference files (e.g. SCENARIOS.md, ISSUES.md) and link to them from the body, trimming SKILL.md to an overview.

De-duplicate the claude --print / hooks-don't-fire rationale so it appears once, then reference it from the DO NOT list.

Replace the '~25s' heuristic wait with a verified checkpoint (e.g. poll until the debug log exists) and confirm each parallel install succeeded before spawning its pane.

Dimension	Reasoning	Score
Conciseness	Operational and largely lean — bash, grep patterns, and tables with minimal concept explanation — but the claude --print prohibition and its rationale are restated across the 'How Evals Work' and 'DO NOT' sections, leaving minor trim opportunities.	4 / 5
Actionability	Copy-paste-ready bash commands cover setup, launch, monitoring, verification, and cleanup, with clearly marked placeholders and a scenario table of explicit prompts and expected skills.	5 / 5
Workflow Clarity	A clear setup→launch→monitor→verify→fix→release→repeat sequence with explicit gates and feedback loops, but a few checkpoints are heuristic rather than verified (the ~25s wait) and the parallel-install loop doesn't confirm each install succeeded.	4 / 5
Progressive Disclosure	No bundle files exist, so all content lives in one ~310-line SKILL.md; the well-sectioned structure helps, but reference-like material (the 12-row scenario table and 15-row issues table) is inlined rather than split into one-level-deep reference files.	3 / 5
	Total	16 / 20 Passed

Description

46%Weight 40%Scale 1-5

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 conveys a clear niche and lists many relevant platform features, but it is missing an explicit 'when to use' trigger and leans on internal jargon rather than natural user keywords. It is specific enough to avoid most conflicts but under-delivers on trigger guidance.

Suggestions

Add an explicit 'Use when ...' clause naming the natural situations that call for this skill (e.g. 'Use when evaluating whether the plugin injects the right skills for a Vercel build').

Replace internal terms like 'skill injection' and 'stress-test' with phrases a user would naturally say, and add synonyms (benchmark, eval, test run, agent evaluation).

Lead with one or two concrete actions ('Launch real Claude Code eval sessions and produce a skill-coverage report') before enumerating platform features.

Dimension	Reasoning	Score
Specificity	Names a broad domain and many platform features but only one to two concrete actions ('benchmark scenarios that push', 'stress-test skill injection'), so coverage of actions is limited rather than comprehensive.	3 / 5
Completeness	Gives a clear 'what' but no 'Use when...' clause or equivalent trigger guidance, which per the rubric caps completeness at 3.	3 / 5
Trigger Term Quality	Mostly internal/technical jargon ('skill injection', 'stress-test', platform feature names) with only one or two generic keywords; it lacks the natural phrases a user would actually say.	2 / 5
Distinctiveness Conflict Risk	The niche (Vercel plugin skill-injection benchmarking) is specific and largely distinct, with only minor overlap risk against general Vercel or eval skills.	4 / 5
	Total	12 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: vercel/vercel-plugin
Path: .claude/skills/benchmark-agents/SKILL.md
Commit: 11c3258

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

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.