mirror of
https://github.com/mukul975/Anthropic-Cybersecurity-Skills.git
synced 2026-07-17 21:19:40 +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.
81 lines
3.0 KiB
Markdown
81 lines
3.0 KiB
Markdown
# garak CLI Reference
|
|
|
|
Source: https://github.com/NVIDIA/garak and https://reference.garak.ai/en/latest/cliref.html
|
|
|
|
## Invocation
|
|
|
|
```bash
|
|
python -m garak <options>
|
|
# or the console script
|
|
garak <options>
|
|
```
|
|
|
|
## Core flags
|
|
|
|
| Flag | Description | Example |
|
|
|------|-------------|---------|
|
|
| `--target_type` (alias `--model_type`) | Generator family / interface | `--target_type openai` |
|
|
| `--target_name` (alias `--model_name`) | Specific model name | `--target_name gpt-4o-mini` |
|
|
| `--probes`, `-p` | Comma-separated probe spec; module or `module.Class` | `--probes promptinject,dan.Dan_11_0` |
|
|
| `--detectors`, `-d` | Override detectors | `--detectors mitigation.MitigationBypass` |
|
|
| `--generations`, `-g` | Completions generated per prompt | `--generations 5` |
|
|
| `--parallel_attempts` | Parallel probe attempts (throughput) | `--parallel_attempts 8` |
|
|
| `--report_prefix` | Prefix for output report files | `--report_prefix baseline` |
|
|
| `--config` | Load a YAML/JSON run config | `--config assessment.yaml` |
|
|
| `-G` / `--generator_option_file` | JSON file with generator options (REST etc.) | `-G rest.json` |
|
|
| `--generator_options` | Inline JSON generator options | |
|
|
| `--list_probes` | Print all probes and exit | |
|
|
| `--list_detectors` | Print all detectors and exit | |
|
|
| `--list_generators` | Print all generator types and exit | |
|
|
| `--list_buffs` | Print prompt-mutating buffs | |
|
|
| `--version` | Print version | |
|
|
| `--verbose`, `-v` | Increase logging | |
|
|
|
|
## Common target_type values
|
|
|
|
| Value | Target |
|
|
|-------|--------|
|
|
| `huggingface` | Local Hugging Face model |
|
|
| `openai` | OpenAI / OpenAI-compatible API (uses `OPENAI_API_KEY`) |
|
|
| `rest` | Arbitrary HTTP endpoint via JSON spec |
|
|
| `ggml` / `nim` / `replicate` / `cohere` / `bedrock` | Other hosted/local backends |
|
|
|
|
## REST generator JSON spec keys
|
|
|
|
| Key | Meaning |
|
|
|-----|---------|
|
|
| `uri` | Endpoint URL |
|
|
| `method` | HTTP method (`post`) |
|
|
| `headers` | Request headers (supports `$ENV_VAR`) |
|
|
| `req_template_json_object` | Request body with `$INPUT` placeholder |
|
|
| `response_json` | `true` if response is JSON |
|
|
| `response_json_field` | JSONPath to extract the model output |
|
|
|
|
## Output artifacts
|
|
|
|
| File | Content |
|
|
|------|---------|
|
|
| `<prefix>.report.jsonl` | Line-delimited record of every attempt, generation, and detector verdict |
|
|
| `<prefix>.report.html` | Human-readable scorecard |
|
|
| `<prefix>.hitlog.jsonl` | Confirmed successful attacks (evidence) |
|
|
| `garak.log` | Debug log |
|
|
|
|
## Example runs
|
|
|
|
```bash
|
|
# Enumerate
|
|
garak --list_probes
|
|
|
|
# Local HF model, single jailbreak probe
|
|
python -m garak --target_type huggingface --target_name gpt2 --probes dan.Dan_11_0
|
|
|
|
# Hosted API, injection + leakage suite
|
|
export OPENAI_API_KEY="sk-..."
|
|
python -m garak --target_type openai --target_name gpt-4o-mini \
|
|
--probes promptinject,latentinjection,leakreplay \
|
|
--generations 5 --parallel_attempts 8 --report_prefix assessment
|
|
|
|
# Arbitrary REST endpoint
|
|
python -m garak --target_type rest -G rest.json --probes promptinject,dan
|
|
```
|