Offensive-security orchestrator — a unified engagement methodology, attack-first persona, and routing layer over security specialists.
An orchestrator skill for AI-assisted IDEs (Claude Code, Claude.ai, etc.) that brings blackhat depth with whitehat rules to security testing, penetration testing, red-teaming, and smart-contract auditing workflows.
Claude Code (Desktop/IDE):
- Clone or download this skill
- Place in
~/.claude/skills/viktor-blackhat/ - Restart Claude Code
- The skill is now available: trigger it with
/viktor-blackhator auto-trigger on security/audit/pentest requests
Claude.ai:
- Navigate to Settings → Skills
- Add custom skill from this repository
- The skill will be available in your sessions
Viktor Blackhat routes you to the right security methodology and specialist skills based on your target:
- Web/API security testing (REST, GraphQL, RPC, HTTP smuggling, cache poisoning)
- Cloud & Container auditing (AWS, GCP, Azure, K8s, IAM, cloud account posture)
- Smart contracts & blockchain (EVM/Solidity, Solana/Anchor, Cosmos, Cairo, Substrate, TON, Algorand)
- Mobile apps (APK/IPA reverse engineering, auth, data storage)
- AI/LLM applications (prompt injection, jailbreaks, confused deputy)
- Native binaries (C/C++, memory safety, reverse engineering, exploit development)
- Source code (SAST, static analysis, code review, differential review)
- Cryptography (algorithm review, TLS/cipher posture, side-channel analysis)
- Fuzzing (harness writing, coverage analysis, parser/codec fuzzing)
- Network & infrastructure (recon, OSINT, initial access, IDS evasion)
- Supply chain & secrets (dependency audit, leaked credential detection)
- Defense & detection (SIEM, threat hunting, incident response, malware analysis)
This skill brings the full depth and intent of people who actually drain systems, pointed only at code you own or are hired to break:
- Attack economics — Does the exploit pay after gas, slippage, and flash-loan fees? Model attacks with borrowed capital + atomicity.
- Real-exploit hunting — Hunt by pattern-matching real on-chain incidents (flash-loan + oracle manipulation, inflation attacks, signature replay, etc.) onto your code.
- Weaponized PoCs — Every finding includes a runnable, profitable attack transaction (Foundry fork, Solana test-validator, invariant fuzzer).
- Profit-first triage — Severity is measured by net value extracted, repeatability, and reachability — not by how scary the pattern looks.
- Money path first — Spend your time on the vault, the mint authority, the upgrade key. Rank every entry point by how much value sits behind it.
Authorized testing only. This skill operates only on systems you own or have explicit written authorization to test. The gate is non-negotiable:
- ✓ Your own code (pre-ship auditing)
- ✓ Code you're hired to audit (bug bounties, contests, private audits)
- ✓ Deliberately-vulnerable practice targets (Damn Vulnerable DeFi, Ethernaut, etc.)
- ✗ Third-party systems without explicit written scope
- ✗ Destructive denial-of-service or data destruction
- ✗ Mass-targeting or operations without consent
- Assume breach. Start from "the perimeter is already past."
- Evidence over assertion. A finding doesn't exist until proven with a concrete trigger/PoC or traced source→sink flow.
- Depth over breadth. One fully-traced exploit chain on the money path beats twenty shallow grep hits.
- Offense paired with remediation. Every finding ships with a fix and detection strategy.
- No hallucination. Never invent tool names, CVE IDs, paths, or line numbers — only cite what you've run or read.
- Reproducible PoC. State exact input, code path, precondition, and observed effect. Defang live payloads; reference source file paths instead.
viktor-blackhat/
├── SKILL.md # Skill entry point (metadata + methodology)
├── references/
│ ├── methodology.md # 6-phase engagement lifecycle
│ ├── persona-and-principles.md # Offensive operator layer, voice rules, auth gate
│ ├── verification-and-reporting.md # Anti-false-positive gates, severity, templates
│ ├── routing-map.md # Target type → playbook → specialist routing
│ └── playbooks/
│ ├── a-web-api.md # Web/REST/GraphQL/RPC security
│ ├── b-cloud-container.md # Cloud, IAM, K8s
│ ├── c-code-review-sast.md # Source code, static analysis
│ ├── d-crypto.md # Cryptography, TLS, protocols
│ ├── e-fuzzing.md # Parser, codec, protocol fuzzing
│ ├── f-blockchain.md # EVM/Solana/Cosmos/Cairo/Substrate (with offensive layer)
│ ├── g-mobile.md # Mobile apps (APK/IPA)
│ ├── h-ai-llm.md # LLM security, prompt injection
│ ├── i-recon-network.md # Recon, OSINT, initial access
│ ├── j-offensive-postex.md # Exploit development, post-exploitation
│ ├── k-supply-chain-secrets.md # Dependencies, secrets, supply chain
│ └── m-defense-detection.md # Blue team, detection, IR
The skill auto-triggers on requests like:
- "audit a smart contract"
- "pentest this API"
- "red-team my cloud account"
- "break this contract to its limits"
- "find DeFi vulnerabilities"
- "security test this app"
/viktor-blackhat
Then describe the target and the skill routes you to the appropriate playbook and specialist skills.
Viktor Blackhat orchestrates 100+ specialist skills including:
Web/API: owasp-audit, web-pentest, api-audit, 09-web-security, insecure-defaults, security-payloads, security-webshells
Cloud: cloud-audit, iam-audit, 10-cloud-security, exploiting-cloud-platforms, secrets-audit
Blockchain: solana-rust-deep-audit, web3-blockchain, entry-point-analyzer, dimensional-analysis, token-integration-analyzer, spec-to-code-compliance, plus per-chain scanners (solana-vulnerability-scanner, cosmos-vulnerability-scanner, cairo-vulnerability-scanner, substrate-vulnerability-scanner, ton-vulnerability-scanner, algorand-vulnerability-scanner)
Code review: owasp-audit, codeql, semgrep, c-review, variant-analysis, differential-review, trailmark-structural
Crypto: crypto-audit, 13-crypto-analysis, constant-time-analysis, wycheproof, zeroize-audit
Fuzzing: security-fuzzing, harness-writing, coverage-analysis, address-sanitizer
Exploitation: 03-exploit-development, vuln-research, 04-reverse-engineering
Post-engagement: fp-check (false-positive gate), finding-triage (disposition + severity), second-opinion (adversarial re-audit)
For web3/smart-contract auditing specifically, see the detailed web3 tool stack:
- EVM/Solidity: Slither, Foundry (forge fuzz + mainnet fork), Aderyn, Echidna, Medusa, Halmos, Certora Prover, Mythril, HEVM
- Solana/Anchor: Trident (stateful fuzzing), cargo-audit, cargo-geiger, solana-lints, proptest
- Exploit PoCs: DeFiHackLabs (real hack reproductions), Foundry fork testing, mainnet-fork PoCs
- Formal verification: Halmos (symbolic execution over Foundry tests), Certora CVL, Kontrol
- Learning/Reference: Cyfrin Updraft, Solodit (vulnerability DB), public audit reports
Target: EVM-based lending protocol (Solidity)
1. Skill routes to f-blockchain.md → EVM/Solidity playbook
2. Methodology (6 phases):
- Recon: Detect languages, layout, dependency versions
- Entry-point map: List all state-changing functions
- Dependency check: cargo audit, npm audit
- Parallel hunt: Account model, DeFi economics, off-chain services
- Dimensional analysis: Annotate arithmetic units (D18{tok}, D6{usd})
- Token integration: Check weird-ERC20 patterns
- Structural analysis: Upgrade authority, proxies, invariants
- Fuzz: Invariant fuzzing for broken economic properties
- PoC: Foundry fork with profit calculation
- Report: Severity-graded findings with fixes
3. Offensive layer:
- Flash-loan + oracle → can I move the price and settle at manipulation?
- Rounding → can I loop the error to drain value?
- First-depositor inflation → can value reach the vault outside accounting?
- Profit = extracted − flash_fee − slippage − gas; if positive → CRITICAL
4. Specialists invoked:
- slither (fast SAST pass)
- echidna / medusa (property-based fuzzing)
- foundry (mainnet fork PoCs)
- halmos (formal verification for key invariants)
- fp-check (kill false positives)
- finding-triage (disposition & severity)
This skill is maintained and extended for your own security testing workflows. To customize:
- Clone the repo
- Edit playbooks or add new references
- Update persona-and-principles.md if you want to adjust operating principles
- Push back to your fork or create a PR
This skill is provided as-is for authorized testing only. Use only on systems you own or have explicit written authorization to test.
Viktor — v1ktorrr0x on GitHub
Built for web3 auditors, security researchers, red teamers, and anyone who wants to break systems they own to make them stronger.
Remember: Blackhat depth, whitehat rules. The authorization gate is not a suggestion. Every finding must be proven before reporting. The goal is a stronger system, not a clever report.