minicpm5-deploy-llama-cpp

Run MiniCPM5-1B with llama.cpp using the released GGUF artifacts (F16 / Q8_0 / Q4_K_M). Use when the user wants CPU-only / consumer-GPU / cross-platform native deployment, asks for "llama.cpp", "llama-cli", "llama-server", "GGUF", or has no Python available.

Quality

88%

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

Quality

Content

90%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.

A concise, highly actionable deployment runbook with a clear sequenced workflow and validation step. The only real gap is the broken external reference link and the absence of an explicit error-recovery loop after validation.

Suggestions

Fix or remove the dead reference link to ../../docs/deployment/llama_cpp.md (file does not exist), or create that doc so the one-level-deep reference resolves.

Add a brief feedback loop after the Validate step (e.g. "If the reply is not '2', recheck NGL/CTX and the model path, then re-run") to lift workflow_clarity to 5.

Optionally split the 'Building your own GGUF (advanced)' section into a referenced file if the skill grows, to keep the main body as a pure overview.

Dimension	Reasoning	Score
Conciseness	Lean runbook of tables and terse code blocks with no padding or explanation of concepts Claude already knows (e.g. no "what is llama.cpp"); every section earns its tokens, matching the 5-anchor.	5 / 5
Actionability	Copy-paste-ready commands cover install (macOS/prebuilt/source), download, CLI chat, HTTP server, validation curl, and custom GGUF build — fully executable across the common cases, matching the 5-anchor.	5 / 5
Workflow Clarity	Clear numbered sequence (1 install → 2 download → 3a/3b run → 4 validate) with an explicit validation checkpoint ("Expected: \"2\" in the reply"), but no error-recovery feedback loop, so it sits at the 4-anchor rather than 5; the destructive/batch cap does not apply since this is a non-destructive run skill.	4 / 5
Progressive Disclosure	Well-organized sections with appropriately inlined runbook content and a signaled cross-skill LoRA pointer, but the single external reference "docs/deployment/llama_cpp.md" is a dead link (file not present), a minor navigation gap that keeps it below 5.	4 / 5
	Total	18 / 20 Passed

Description

87%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.

A strong, concrete description that clearly states both the capability and explicit natural-language triggers, tightly scoped to a distinct niche. Minor room to add common synonyms for a perfect trigger-term spread.

Dimension	Reasoning	Score
Specificity	Quotes "Run MiniCPM5-1B with llama.cpp using the released GGUF artifacts (F16 / Q8_0 / Q4_K_M)" — one concrete action populated with specific model/runtime/quant details; above the 3-anchor (1-2 concrete actions) but not the 5-anchor's multiple distinct actions.	4 / 5
Completeness	Explicitly answers both: what ("Run MiniCPM5-1B with llama.cpp using the released GGUF artifacts") and when ("Use when the user wants CPU-only / consumer-GPU / cross-platform native deployment, asks for...") with concrete trigger phrases, matching the 5-anchor.	5 / 5
Trigger Term Quality	Quotes natural triggers "llama.cpp", "llama-cli", "llama-server", "GGUF", plus "CPU-only / consumer-GPU / cross-platform native deployment" and "no Python available" — good coverage with a few common synonyms (e.g. "local LLM", "run locally") missing, so not a 5.	4 / 5
Distinctiveness Conflict Risk	Niche is narrowly scoped to MiniCPM5-1B + llama.cpp GGUF deployment with distinct, specific triggers — minimal overlap risk with sibling deploy skills, matching the 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-llama-cpp/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.