SKILLEMALL.ai
GitHub ↗ RU Sign in

BB ai-guardian

Use this skill whenever the user needs to observe or govern on-endpoint local LLMs running on Ollama, llama.cpp (llama-server), LM Studio, or a local single-node vLLM — inventory installed/running models with an allow/deny verdict (shadow-AI detection), inspect VRAM residency, model license/params/capabilities and server version, view the model policy, detect model provenance/digest drift (re-pulled or tampered weights; strong for Ollama/llama.cpp, id-only and honestly weaker for LM Studio/vLLM), scan a prompt for secrets / PII / source-code / jailbreak with a weighted risk band, route a prompt THROUGH a guard that scans + policy-gates + records + runs-if-allowed (guarded_generate / observe_chat), query the observed-usage log, and roll up anomalies (shadow models, digest drift, high-risk + blocked prompts). Always use this skill for "what local models are installed", "find shadow / unsanctioned AI models", "which model is loaded in VRAM", "scan this prompt for secrets/PII before sending", "stop secrets leaking into a local model", "block a prompt with an API key", "detect a jailbreak / prompt injection", "set a model allowlist / denylist", "detect a tampered / re-pulled model", "audit local LLM usage", "guard my llama.cpp / LM Studio / local vLLM endpoint", or "the complement to IGEL AI Armor". Do NOT use for GPU inference CLUSTERS (multi-node / fleet-scale vLLM / Ray serving) — this is for single-endpoint LOCAL LLMs; point cluster/serving work to inference-aiops. Also not for hypervisors, storage, backup, Kubernetes, or network devices. Passive inventory/state auditing plus opt-in route-through content governance, with a bundled governance harness (audit, policy, token budget, undo, risk-tiers). A transparent capture proxy is v0.2 roadmap.

ClawHub Claude Code author: wei zhou v0.11.2 MIT-0 6 files body ≈ 2 816 tokens Open the sourceclawhub.ai↗ analyzed 10 h ago

cpp (llama-server), LM Studio, or a local single-node vLLM — inventory installed/running models with an allow/deny verdict (shadow-AI detection), inspect VRAM…

As a process B 71/100 · Nearly there — weak spots: result and completion, inputs and preconditions

ProcedureKubernetesAI and agentstype and topics are labelled automatically from the skill text
Runs in: OpenClaw Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) Goose OpenCode
JSON
Technical rating
B
83/100
safety, quality, tests
Safety 60%
95
Quality 40%
64
Run on models
none yet
Process rating
B
71/100
Nearly there
Result and completion w 14
0
Inputs and preconditions w 11
0
Failures and branches w 10
55
the three weakest of ten parameters · all ten

What is at stake

Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.

Broad scope medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The skill asks for more than the task needs: broad tool access, credential environment variables, binaries. Every extra permission widens the damage from a mistake or a compromise.

For the author

Narrow allowed-tools and the variable list to the minimum; replace binaries with readable sources or scripts.

How to improve

  1. Shorten the description to 1024 characters.
For the model run — optional
  • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
  • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

Guard findings · 1

✓ No critical or high findings

Medium and low: 1
  • medium Broad scope meta-broad-allowed-tools SKILL.md:1
    Broad tool permissions pre-approved: Bash
    allowed-tools: Bash

Files scanned: 6. Evidence is masked. Grey chips explain why severity was lowered.

Against the Agent Skills spec

  • error description-long description is 1771 chars, limit 1024
  • note description-budget description takes 1771 of the ~15000-char shared budget for all skills
  • note frontmatter-key unknown frontmatter key "slug"
  • note frontmatter-key unknown frontmatter key "displayName"
  • note frontmatter-key unknown frontmatter key "summary"
  • note frontmatter-key unknown frontmatter key "homepage"
  • note frontmatter-key unknown frontmatter key "installer"

Process rating: all ten parameters 71/100

  • 0Result and completion. Does not say what the result is
  • 0Inputs and preconditions. Does not say what the process needs to start
  • 55Failures and branches. 1 branches
  • 100Tools and files. Tools declared in frontmatter
  • 100Steps. 41 steps
  • 100When it triggers. States when to use and when not to
  • 100Consistency. Name and required fields are in place
  • 100Execution cost. Instruction body is 2816 tokens
  • 100Running it twice. Mutating operations check current state
  • 100Progress reporting. Reports progress
  • low The response is described with custom markup (4 tags): a typed call is more reliable

Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

Quality signals

  • +3Description length 1770: 120–800 characters recommended
  • +3Output format is not stated: the model decides each time
  • +2Single-language instructions
  • +5Description quotes 12 example trigger phrases
  • +4Description says when NOT to use the skill
  • +4Structure: 12 headings
  • +3Step-by-step instructions: 41 items
  • +4Has examples (2 code blocks)
  • +4Reference files are cited in the instructions (3 of 4)
  • +1License stated

Quality base 70; lint remarks subtract, signals add up to 100. Result: 64.

External checks

ClawHub: suspicious
This skill is not clearly malicious, but it asks for local model administration powers while leaving important install and authorization controls unclear.
LLM: suspicious (high) · 12 Sept 2026