llm-integration
Expert skill for integrating local Large Language Models using llama.cpp and Ollama. Covers secure model loading, inference optimization, prompt handling, and protection against LLM-specific vulnerabilities including prompt injection, model theft, and denial of service attacks.
79
76%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./skills/llm-integration/SKILL.mdQuality
Discovery
67%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This description demonstrates strong specificity with concrete actions and named tools, and occupies a distinct niche. However, it lacks an explicit 'Use when...' clause which limits its effectiveness for skill selection, and could benefit from more natural trigger terms that users would actually say when needing this skill.
Suggestions
Add a 'Use when...' clause with explicit triggers like 'Use when setting up local LLM inference, configuring Ollama or llama.cpp, or securing self-hosted AI models'
Include more natural user terms like 'run models locally', 'self-hosted AI', 'gguf files', 'quantized models', 'local inference'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: 'model loading', 'inference optimization', 'prompt handling', and protection against specific vulnerabilities ('prompt injection, model theft, denial of service attacks'). Names specific tools (llama.cpp, Ollama). | 3 / 3 |
Completeness | Clearly answers 'what' (integrating local LLMs with specific tools and security concerns), but lacks an explicit 'Use when...' clause or equivalent trigger guidance. The 'when' is only implied. | 2 / 3 |
Trigger Term Quality | Includes some good technical terms users might say ('llama.cpp', 'Ollama', 'LLM', 'prompt injection'), but missing common variations like 'local AI', 'run models locally', 'self-hosted LLM', 'gguf', 'quantized models'. | 2 / 3 |
Distinctiveness Conflict Risk | Clear niche focusing specifically on local LLM integration with named tools (llama.cpp, Ollama) and security concerns. Unlikely to conflict with general coding skills or cloud-based AI integration skills. | 3 / 3 |
Total | 10 / 12 Passed |
Implementation
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured skill with excellent actionability and progressive disclosure. The code examples are executable and security-focused, with clear workflows and validation checkpoints. The main weakness is moderate verbosity in introductory sections that explain concepts Claude already understands, though the technical content itself is appropriately dense.
Suggestions
Trim Section 1 Overview and Section 2 Core Principles - remove explanatory text about what prompt injection is or why TDD matters; Claude knows these concepts
Condense Section 3 Core Responsibilities into a brief bullet list rather than prose explanations of security principles
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill contains some unnecessary verbosity in the overview and principles sections that Claude already knows (e.g., explaining what prompt injection is, general TDD principles). However, the code examples are reasonably efficient and the tables are well-condensed. | 2 / 3 |
Actionability | Provides fully executable Python code examples with proper imports, concrete configuration classes, and copy-paste ready implementations. The patterns include specific version numbers, exact commands, and working code snippets. | 3 / 3 |
Workflow Clarity | The TDD workflow in Section 6 provides clear sequencing with explicit steps (write test → implement → refactor → verify). The pre-deployment checklist provides validation checkpoints, and security patterns include explicit verification steps like checksum validation. | 3 / 3 |
Progressive Disclosure | Excellent structure with clear overview in SKILL.md and well-signaled one-level-deep references to `references/advanced-patterns.md`, `references/security-examples.md`, and `references/threat-model.md`. Each section appropriately points to detailed materials without nesting. | 3 / 3 |
Total | 11 / 12 Passed |
Validation
68%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 16 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
skill_md_line_count | SKILL.md is long (609 lines); consider splitting into references/ and linking | Warning |
description_trigger_hint | Description may be missing an explicit 'when to use' trigger hint (e.g., 'Use when...') | Warning |
metadata_version | 'metadata' field is not a dictionary | Warning |
license_field | 'license' field is missing | Warning |
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 11 / 16 Passed | |
- Repository
- martinholovsky/claude-skills-generator
- Commit
1086ef2
- 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.