minicpm5-deploy-vllm

Serve MiniCPM5-1B or MiniCPM5-2B via vLLM as an OpenAI-compatible HTTP server. Use when the user wants high-throughput production serving on NVIDIA GPU, asks for "vLLM", "OpenAI server", "REST API for MiniCPM5", or "production deployment".

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

88%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 with fully executable install/launch/validate commands, a clear sequenced workflow with validation, and good section organization. The main weaknesses are time-sensitive version/date detail in the Tool calling section and a body length that slightly exceeds the simple-skill threshold for top progressive-disclosure marks.

Suggestions

Move the time-sensitive specifics in the Tool calling section (PR #43175, merge date 2026-05-27, v0.22.0) into a short 'Status / deprecation' note or a separate reference so they do not penalize conciseness as they age.

Consider extracting the Tool calling plugin subsection into its own reference file (e.g. tool-calling.md) and linking to it from the main body, which would tighten the core workflow and improve progressive disclosure.

Add an explicit 'if validation fails, do X' feedback loop to the Validate step (e.g. check the curl response code, then consult Common pitfalls) to make the recovery path part of the workflow rather than a separate section.

Dimension	Reasoning	Score
Conciseness	Lean and assumes Claude's competence throughout (no explanations of what vLLM or GPUs are), with every section earning its place; held below 5 because the Tool calling section embeds time-sensitive specifics (PR #43175, merge date 2026-05-27, v0.22.0) outside a deprecated/old-patterns section, which the guidelines penalize.	4 / 5
Actionability	Fully executable, copy-paste-ready commands for install, launch, and validate, plus the tool-calling variant, with a concrete validation curl and an explicit expected output; specific examples cover the common cases, matching the score-5 anchor.	5 / 5
Workflow Clarity	Clear numbered sequence (Install -> Launch -> Validate) with an explicit validation checkpoint ('Wait for Application startup complete', the validate curl, expected output) and error-recovery guidance in Common pitfalls; not a destructive/batch operation so no cap applies.	5 / 5
Progressive Disclosure	Well-organized with clearly signaled sections and a single one-level-deep reference (docs/deployment/vllm.md); not 5 because the body exceeds the ~50-line simple-skill threshold and the Tool calling plugin section is substantial enough that it could be split into a separate reference file.	4 / 5
	Total	18 / 20 Passed

Description

82%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 is specific, third-person, and clearly answers both what the skill does and when to invoke it, with well-chosen natural trigger terms and minimal conflict risk. The only weakness is that it names essentially one serving action rather than multiple distinct capabilities, capping specificity at 3.

Dimension	Reasoning	Score
Specificity	Names the domain and one concrete action ('Serve MiniCPM5-1B or MiniCPM5-2B via vLLM as an OpenAI-compatible HTTP server'), but does not enumerate multiple distinct actions; not below 2 because the action is specific rather than generic, and not 4 because there are not 'several specific actions'.	3 / 5
Completeness	Explicitly answers both 'what' (serve MiniCPM5 via vLLM as an OpenAI-compatible HTTP server) and 'when' (Use when... high-throughput production serving on NVIDIA GPU, asks for 'vLLM', 'OpenAI server', 'REST API for MiniCPM5', or 'production deployment') with concrete trigger phrases, matching the score-5 anchor.	5 / 5
Trigger Term Quality	Good coverage of natural trigger terms a user would say ('vLLM', 'OpenAI server', 'REST API for MiniCPM5', 'production deployment', 'high-throughput production serving'); not 5 because a few natural synonyms are missing, and not 3 because the keywords are well-chosen rather than merely 'some relevant keywords'.	4 / 5
Distinctiveness Conflict Risk	Clear niche (MiniCPM5 served via vLLM on NVIDIA GPU) with distinct triggers and minimal conflict risk; the body's 'When NOT to use' section further disambiguates it from sibling minicpm5-deploy-* skills.	5 / 5
	Total	17 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 1 suspicious	Warning

	Total	15 / 16 Passed

Repository: OpenBMB/MiniCPM
Path: skills/minicpm5-deploy-vllm/SKILL.md
Commit: 316cfb1

Reviewed: 15 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.