minicpm5-deploy

Pick the right inference backend for a MiniCPM5-1B checkpoint and route to a backend-specific cookbook skill. Use when the user wants to deploy / serve / chat-with / benchmark a MiniCPM5 model and has not yet committed to a specific engine, or when they say "deploy MiniCPM5", "run MiniCPM5", "serve MiniCPM5", "MiniCPM5 推理", "部署 MiniCPM5".

Quality

93%

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

96%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 content is a well-engineered router skill: lean, highly actionable, with a clear sequenced workflow and an explicit sanity-check feedback loop. The only gap is progressive disclosure — the body points to external cookbook files that are not present in the bundle.

Suggestions

Add the referenced cookbook files under docs/deployment/ (or include a references/ directory) so the one-level-deep links in section 6 resolve to real files.

If the paired sub-skills are part of the same bundle, list them explicitly in a 'Related skills' section so the routed targets are verifiable as real artifacts.

Consider moving the universal sanity-check curl and the cross-backend pitfalls into a shared reference file referenced from SKILL.md to further trim the overview once the cookbooks exist.

Dimension	Reasoning	Score
Conciseness	The body is lean and assumes Claude's competence — dense tables and exact flags with no tutorial padding about what backends or quantization are — so nearly every token earns its place, matching the level-5 anchor.	5 / 5
Actionability	It provides a concrete decision matrix mapping user intent to a named skill, a copy-paste-ready curl sanity check, and exact CLI flags (e.g. --max-model-len 131072, -c 131072), matching the fully-executable level-5 anchor.	5 / 5
Workflow Clarity	The flow is clearly sequenced (gather input → pick backend → invoke sub-skill → sanity check → pitfalls) with an explicit validation checkpoint and a recovery feedback loop ('If you get <think>... set enable_thinking: false'), matching the level-5 anchor with error-recovery guidance.	5 / 5
Progressive Disclosure	The file itself is well-structured with numbered sections and appropriately delegates detail to sibling sub-skills rather than inlining, but the referenced bundle materials (the ../../docs/deployment/ cookbooks) do not exist in the bundle, so the one-level-deep references cannot be verified, keeping it just below level 5.	4 / 5
	Total	19 / 20 Passed

Description

90%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 strong: it cleanly states a narrow purpose, gives an explicit 'Use when' clause, and lists diverse natural trigger terms including bilingual variants. The only weaker dimension is specificity, which names the action but not a comprehensive list of capabilities.

Dimension	Reasoning	Score
Specificity	The description names the domain and one-two concrete actions ('Pick the right inference backend', 'route to a backend-specific cookbook skill') but does not enumerate a comprehensive set of concrete capabilities, matching the anchor for 1-2 concrete actions without full coverage.	3 / 5
Completeness	It explicitly answers both what (pick backend, route to skill) and when via an explicit 'Use when...' clause with concrete trigger phrases, matching the level-5 anchor that requires both an explicit what and a concrete when.	5 / 5
Trigger Term Quality	It provides comprehensive natural trigger phrases users would actually say — 'deploy MiniCPM5', 'run MiniCPM5', 'serve MiniCPM5', plus non-English variants 'MiniCPM5 推理', '部署 MiniCPM5' — giving broad synonym coverage comparable to the level-5 anchor.	5 / 5
Distinctiveness Conflict Risk	The niche is tightly scoped to MiniCPM5-1B backend selection before an engine is committed, with model-name-specific triggers that make triggering for the wrong skill unlikely, matching the clear-niche level-5 anchor.	5 / 5
	Total	18 / 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/SKILL.md
Commit: 719e4fc

Reviewed: 17 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.