Borrowing it
Nothing to install: this file belongs to aveproject/ave. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/aveproject/ave/main/.claude/skills/research-new-attack-classes/SKILL.mdgit clone --depth 1 https://github.com/aveproject/aveWrote 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/aveproject/ave/research-new-attack-classes)<a href="https://agentmods.dev/skills/aveproject/ave/research-new-attack-classes"><img src="https://agentmods.dev/badge/skills/aveproject/ave/research-new-attack-classes/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/aveproject/ave/research-new-attack-classes"><img src="https://agentmods.dev/badge/skills/aveproject/ave/research-new-attack-classes.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Anti-Refusal · line 40 Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.Fix: Remove instructions that suppress warnings, disclaimers, or ethical commentary. Let the agent surface safety-relevant caveats to the user.
- medium Excessive Agency · line 221 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00000 | $0.02730 |
| Opus 5.5 | $0.00000 | $0.01092 |
| Sonnet 5.5 | $0.00000 | $0.00546 |
| Haiku 4.5 | $0.00000 | $0.00273 |
Grade A, and why
research-new-attack-classes scanned grade A with 0 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 278 lines — stays where its author put it; the contents beside it link to each section on GitHub.
research-new-attack-classes
Keep AVE current with the real threat landscape. This skill researches agentic AI / MCP attack classes from primary sources, benchmarks each against the existing AVE record set, and for every class decides one of three outcomes: ALREADY COVERED, VARIANT OF EXISTING, or NEW CLASS.
For NEW CLASS only, it opens a GitHub issue and hands off to add-ave-record.
This skill never pads. It maps to distinct behavioral classes that exist in cited research. If a "new" attack is just a textual variant of a class AVE already covers, it is logged as a variant, not a new record.
When to run this
- Monthly cadence (the field moves fast — set a recurring reminder)
- After a major disclosure (a new CVE, a new OX/Invariant/HiddenLayer report)
- After a new academic taxonomy or benchmark paper drops
- Before a release (close the gap deliberately)
Do NOT run this to hit a record-count target. Run it to stay current. There is no quota. The right number of records equals the number of distinct behavioral classes that exist — no more.
Inputs
- The current record set in
records/*.json - The schema in
schema/ave-record.schema.json - LANGUAGE.md (attack_class vocabulary, behavioral_fingerprint discipline)
- docs/adr/0001-behavioral-fingerprints.md (the no-signatures rule)
The pipeline — five phases
Phase 1: Research (gather, do not judge yet)
Search primary sources only. Rank by authority:
- MITRE ATLAS (atlas.mitre.org) — the AI-specific ATT&CK. The most authoritative existing enumeration of adversarial AI tactics and techniques. Treat it as BOTH a primary source AND a benchmark target. Capture the ATLAS technique ID (AML.Txxxx) for every candidate that maps to one — these IDs go in the AVE record for cross-referencing.
- CVE/NVD entries and vendor disclosures (OX Security, Invariant Labs, HiddenLayer, Unit 42, Snyk, Trail of Bits)
- Peer-reviewed / arXiv taxonomies and benchmarks (MCPSecBench, MCP-SafetyBench, MCPTox, Formal Security Framework, Hou et al. lifecycle taxonomy, parasitic toolchain studies)
- OWASP updates (MCP Top 10, Agentic AI Top 10, AIVSS, AIBOM)
- Microsoft / Google / NSA / NIST taxonomy and guidance updates (note: ATT&CK proper for any technique that crosses into conventional TTPs)
- Framework changelogs and security advisories (MCP spec, OpenClaw, etc.)
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 · 278 lines · 0 tokens per session scan A e47bb3bfc39a
research-new-attack-classes is a skill published in the GitHub repository aveproject/ave (24 stars, last pushed 2d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 2,730 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
skill-inspector
Review AI agent skills before installation using NVIDIA SkillSpector and source-aware semantic review. Use when asked whether a skill or downloaded skill folder is safe, trustworthy, installable, over-permissioned, or malicious.
stock-report
Prepare a read-only availability report from an identified stock snapshot. Use when a task asks whether a stated quantity is available, without reserving or ordering stock.
crossval-harness
Orchestrate a static + dynamic, exploit-validated red-team of an AI agent — read the source to find candidate vulnerable paths, then run the dynamic skills to confirm or refute each one empirically. The arbiter of truth is whether the exploit works, not a model vote. Authorized testing of agents you own or are…
redteam-an-agent
The end-to-end methodology for red-teaming a specific AI agent — adaptively, exploit-validated, and honestly. Read THIS target's own code, stand up a disposable harness, and prove or refute each weakness through a real attacker-reachable entry point. This is the orchestration + discipline that makes a finding…
audit-agent-code
Security-audit an AI agent's OWN framework code for classic appsec vulnerabilities the LLM can't defend — path traversal, command injection, SSRF, fail-open security controls, missing/late auth, unsafe deserialization — especially at the untrusted-input boundaries (channels, file/media handlers, config & skill…
rt6-memory-poisoning
Red-team an AI agent's persistent memory for cross-session prompt injection ("memory poisoning") — does untrusted content the agent processes get written into long-term memory and re-fire in future sessions with no attacker present? Authorized testing of agents you own or are permitted to test.