mirror of
https://github.com/mukul975/Anthropic-Cybersecurity-Skills.git
synced 2026-06-26 19:54:37 +03:00
mukul975
8cae0648ec
Demand-driven expansion targeting the fastest-growing 2025-2026 threat and
skills categories (ISC2/WEF/CrowdStrike/Mandiant signals):
- AI Security (NEW domain, 12 skills): LLM red-teaming with garak/PyRIT,
prompt injection (direct/indirect/RAG), MCP tool-poisoning, agentic tool
invocation, guardrails, model/data poisoning, system-prompt leakage,
embedding/vector weaknesses, model extraction, continuous red-teaming
- Supply Chain Security (NEW domain, 5 skills): SBOMs, dependency confusion,
malicious-npm triage, typosquatting, SLSA/Sigstore provenance
- Hardware & Firmware Security (NEW domain, 4 skills): CHIPSEC/UEFI audit,
Secure Boot bypass, TPM measured-boot attestation, ESP bootkit hunting
- Identity (10): Entra ID/ROADtools, GraphRunner, AADInternals, ADCS/Certipy,
shadow credentials, coercion, BloodHound CE, device-code phishing, SSO abuse
- Cloud-native (8): Stratus, Pacu, CloudFox, container escape, K8s RBAC,
Falco, Trivy, kube-bench
- Offensive C2 (6): Sliver, Havoc, NetExec, DPAPI, NTLM relay ESC8, redirectors
- DFIR (6): Hayabusa, Chainsaw, KAPE, Velociraptor, EZ Tools, Plaso
- Backfill (4): OpenCTI, MISP, honeytokens, post-quantum crypto migration
Each skill follows the repo taxonomy (SKILL.md + references/{standards,api-reference}.md
+ scripts/agent.py + LICENSE), with researched real tool commands (no placeholders),
complete frontmatter, and ATT&CK/ATLAS + NIST CSF mappings. Updates README domain
table, skill count, and index.json.
51 lines
2.0 KiB
Markdown
51 lines
2.0 KiB
Markdown
# API Reference — Indirect Prompt Injection Detection
|
|
|
|
## LLM Guard
|
|
|
|
Install: `pip install llm-guard`
|
|
|
|
| API | Description |
|
|
|-----|-------------|
|
|
| `from llm_guard.input_scanners import PromptInjection` | Import the injection scanner |
|
|
| `PromptInjection(threshold=0.5, match_type=MatchType.FULL)` | Construct scanner (FULL or SENTENCE match) |
|
|
| `scanner.scan(text)` | Returns `(sanitized_text, is_valid, risk_score)` |
|
|
| `from llm_guard import scan_prompt` | Run multiple scanners over a prompt |
|
|
|
|
`is_valid == False` indicates an injection was detected.
|
|
|
|
## Transformers detector models
|
|
|
|
Install: `pip install transformers torch`
|
|
|
|
| API | Description |
|
|
|-----|-------------|
|
|
| `pipeline("text-classification", model=...)` | Load a classifier pipeline |
|
|
| `protectai/deberta-v3-base-prompt-injection-v2` | Open prompt-injection classifier (labels: SAFE / INJECTION) |
|
|
| `meta-llama/Llama-Prompt-Guard-2-86M` | Meta jailbreak/injection classifier (gated license) |
|
|
|
|
## Content extraction
|
|
|
|
| API | Description |
|
|
|-----|-------------|
|
|
| `BeautifulSoup(html, "html.parser")` | Parse HTML |
|
|
| `soup.find_all(string=lambda t: isinstance(t, Comment))` | Extract HTML comments |
|
|
| `pypdf.PdfReader(path).pages[i].extract_text()` | Extract PDF text |
|
|
| `pytesseract.image_to_string(Image.open(path))` | OCR text from an image |
|
|
| `PIL.Image.open(path)._getexif()` | Read EXIF metadata |
|
|
|
|
## Normalization helpers
|
|
|
|
| Technique | Method |
|
|
|-----------|--------|
|
|
| Strip zero-width chars | `str.translate` over U+200B..U+FEFF |
|
|
| Strip Unicode tag chars | filter ord in range 0xE0000-0xE007F |
|
|
| Canonicalize | `unicodedata.normalize("NFKC", text)` |
|
|
| Decode Base64 | `base64.b64decode(token)` |
|
|
| Decode ROT13 | `codecs.decode(text, "rot_13")` |
|
|
|
|
## External References
|
|
|
|
- LLM Guard PromptInjection docs: https://llm-guard.com/input_scanners/prompt_injection/
|
|
- Hugging Face transformers pipelines: https://huggingface.co/docs/transformers/main_classes/pipelines
|
|
- pytesseract: https://github.com/madmaze/pytesseract
|