Credits

Claude-BugHunter is the author’s bug-hunting stack, with a vendored foundation layer from upstream community work. Full attribution below.

What this repo redistributes

This is a bundle: skills and commands are vendored directly into skills/ and commands/ so install is a single step. Vendored content retains its original license; the original work in this repo (the majority — see breakdown below) is MIT-licensed by the author.

Category Count Source
Original / personally-curated skills 43 This repo
Community-contributed skills (v3) 20 community PRs (e.g. #7)
Vendored foundation skills 8 shuvonsec/claude-bug-bounty (MIT)
Vendored slash commands 15 shuvonsec/claude-bug-bounty (MIT)
Total 71 skills + 15 commands  

Original work in this repo

24 per-class hunt-* skills — curated from disclosed HackerOne reports and engagement data

Each hunt-* skill codifies detection patterns, payloads, and chain templates derived from real disclosed HackerOne reports (21 skills) plus three additional skills (hunt-aspnet, hunt-sharepoint, hunt-ntlm-info) built from authorized engagements involving on-prem SharePoint farms. The selection of report sets, the curation of what to extract, and the resulting skill content are the author’s work, with content derived from publicly disclosed bug-bounty reports (HackerOne’s public disclosures are intended for community learning) and authorized-engagement observations.

The shuvonsec/public-skills-builder generator tool was used as scaffolding to produce skill files from the curated report sets — the tool is acknowledged as inspiration/scaffolding (see “Tooling” below), but the content is the author’s curation work.

Skill H1 reports curated
hunt-misc 225
hunt-xss 174
hunt-rce 67
hunt-idor 26
hunt-subdomain 11
hunt-csrf 10
hunt-oauth 10
hunt-ssrf 9
hunt-sqli 8
hunt-business-logic 7
hunt-cache-poison 4
hunt-auth-bypass 4
hunt-xxe 4
hunt-graphql 3
hunt-race-condition 3
Total disclosed reports curated 574+

Plus 12 additional hunt-* skills curated by topic without an explicit report-count tag: hunt-saml, hunt-ato, hunt-mfa-bypass, hunt-http-smuggling, hunt-ssti, hunt-file-upload, hunt-api-misconfig, hunt-cloud-misconfig, hunt-llm-ai, plus three engagement-derived (hunt-aspnet, hunt-sharepoint, hunt-ntlm-info), plus alternates (hunt-cache-poison, hunt-race-condition, hunt-subdomain), plus the meta-router hunt-dispatch.

Other personal skills

  • offensive-osint (v3.0) — Refactored from a 4,168-line monolith into a lean SKILL.md (~400 lines) plus 15 modular reference files in references/ (subdomain enum, identity fabric, secret patterns, dorks, sector-specific recon, etc.). Detail content loads on demand — Claude reads only the relevant references for the current task.
  • osint-methodology (v2.1) — 5-stage recon pipeline, 29-type asset graph, severity rubric, identity-fabric mapping, vulnerability prioritization (CVE/EPSS/KEV), bug bounty submission templates, threat-actor investigation, cryptocurrency tracing, image/video forensics.
  • bugcrowd-reporting — Bugcrowd-specific reporting tactics: VRT category fallback hierarchy, severity-request paragraphs, OOS-clause rebuttal templates (rate limiting on auth-flow endpoints, debug-info framing, user-enumeration with sensitive PII, theoretical-issue counter), chained-finding cross-reference patterns, target selection for QA-vs-prod programs, researcher-side hygiene.
  • evidence-hygiene — Cookie redaction protocols, PII black-bar discipline, HAR sanitization recipes, Burp/DevTools screenshot patterns, post-submission rotation hygiene. The redaction protocol distinguishes “your-account secrets” (always redact) from “other-user PII” (redact-by-default with explicit cross-account-impact exception) from “triager-useful metadata” (leave visible).
  • bb-local-toolkit — Personal customization of the master bug-bounty workflow with author’s pipeline preferences.

Enterprise-platform attack skills

Built from authorized red-team engagements (enterprise targets including on-prem SharePoint farms) plus public CVE / advisory catalogues and IdP vendor documentation. Each skill is original work — vendor docs and public CVEs provided the technical primitives; the curation, current 2024-2026 chain assembly, and operator-discipline framing are the author’s.

  • m365-entra-attack — M365 / Entra ID full chain. AADSTS error reference, user enum vectors (with hardening status), Smart Lockout math, Conditional Access bypass options, ROPC + SAML SSO browser flow. ROPC spray surfaced pre-existing lockouts and CA-blocked credentials during authorized work.
  • okta-attack — Okta-as-IdP attack chain for orgs where Okta sits alongside or instead of Entra. Distinct endpoints, distinct rate-limiting, distinct factor flows.
  • cloud-iam-deep — AWS / Azure / GCP IAM red-team post-credential model. 24+ AWS, 8+ Azure, 6+ GCP priv-esc patterns. Built for the “recon yielded a credential, what does it grant” workflow.
  • vmware-vcenter-attack — vSphere / vCenter / Workspace ONE / Aria external attack matrix. Internet-exposed only.
  • enterprise-vpn-attack — Cisco ASA, Fortinet, Citrix NetScaler, PAN GlobalProtect, Pulse/Ivanti, SonicWall, F5 — versioning, CVE matrix 2018-2026, AAA backend identification, default credentials, config-disclosure paths.
  • apk-redteam-pipeline — End-to-end Android APK pipeline. Multiple APKs processed manually during authorized work; hardcoded JWT + internal API endpoints recovered.
  • supply-chain-attack-recon — Recon and identification ONLY — actual package publishing / typosquat attacks require explicit written sign-off because they can affect entire npm/PyPI ecosystems.
  • hunt-sharepoint — SharePoint Server 2013–Subscription Edition on-prem farms. Anonymous endpoint enum, legacy SOAP login bypass, ToolShell precondition chain (CVE-2025-53770), SafeControl reflection enumeration, NTLM Type-2 disclosure, custom-zone Forms auth bridging. Built from authorized engagement against an EoL SharePoint farm.
  • hunt-aspnet — ASP.NET-specific surface. ViewState deserialization, machineKey recovery, dual-parser MAC-bypass anti-pattern, request-validator bypass. Same SharePoint engagement.
  • hunt-ntlm-info — NTLM/Negotiate anonymous information disclosure on internet-reachable IIS/SharePoint/Exchange. AV_PAIRS leakage of internal DNS forest, NetBIOS domain, computer name, AD timestamp. Same SharePoint engagement.

Red-team tradecraft skills

  • redteam-mindset — Operator discipline corrections that separate offensive red-team work from defensive WAPT. Load at start of every red-team engagement; reload whenever feeling stuck on a defended target.
  • mid-engagement-ir-detection — Methodology for detecting client SOC patches, attacker activity, and security-state changes that occur DURING a red-team engagement. Built after observing a client patch a confirmed SQLi within 30 minutes of detection AND an external attacker lock 14 new accounts during a single test session.
  • redteam-report-template — Client-facing deliverable format: Subject / Observations / Description / Impact / Recommendation / PoC. Built from a 14-finding deliverable (52KB MD + 2.2MB DOCX with 16 embedded screenshots).

Tooling and docs

  • hunt <target> shell command — Engagement-folder scaffolding: creates ~/Targets/<name>/ with CLAUDE.md, scope.md, findings/, evidence/, submissions.txt, notes.md, and a sensible .gitignore for engagement artifacts.
  • Bundle packaging — Single-step installer (scripts/install.sh) that copies all 71 skills, 15 commands, and the hunt scaffold into ~/.claude/.
  • assets/banner-v2.svg — Hand-coded SVG banner.
  • Documentation — README.md, INSTALL.md, USAGE.md, CONTRIBUTING.md, docs/architecture.md, this credits file.

Vendored foundation (from shuvonsec/claude-bug-bounty)

These 8 skills + 15 slash commands form the methodology backbone of the bundle. Vendored as-is (MIT-licensed) so the entire stack installs in one step.

Skills (8)

Skill Purpose
bb-methodology 5-phase non-linear hunting workflow + critical-thinking framework
bug-bounty Master orchestrator
triage-validation 7-Question Gate, 4 pre-submission gates, never-submit list
report-writing H1 / Bugcrowd / Intigriti / Immunefi report templates, CVSS 3.1 + 4.0
security-arsenal Payloads, bypass tables, wordlists, gf patterns
web2-recon Subdomain enumeration, host discovery, URL crawling
web3-audit 10 DeFi bug classes, Foundry PoC template
meme-coin-audit Token rug-pull detection

Slash commands (15)

/hunt /recon /scope /triage /validate /report /autopilot /chain /intel /pickup /surface /remember /memory-gc /token-scan /web3-audit

Repo: https://github.com/shuvonsec/claude-bug-bounty License: MIT (verify in upstream repo)

Tooling acknowledgments (not vendored — used as scaffolding)

shuvonsec — public-skills-builder

Generator tool that produces skill scaffolding from disclosed HackerOne reports. Used to generate the initial scaffolds for the per-class hunt-* skills before the author’s curation. The tool itself is not redistributed in this repo.

Repo: https://github.com/shuvonsec/public-skills-builder License: MIT (verify in upstream repo)

Inspirations

archangel / douglasday

A top-10 historical HackerOne hunter. The per-class hunt-* pattern with chain templates from disclosed reports was inspired by his public skill stack screenshots, plus the hunt <target> engagement-scaffolding shell pattern.

Trail of Bits — trailofbits/skills

Skill-authoring discipline reference. Their CLAUDE.md states:

“Skills should be specific and actionable rather than reference dumps, focusing on behavioral guidance over comprehensive documentation.”

This principle informed the offensive-osint v3 refactor (lean SKILL.md + references/ subfolder for progressive disclosure).

Repo: https://github.com/trailofbits/skills

SecSkills — trilwu/secskills

16 specialized security skills + 6 expert AI subagents. Demonstrated the subagent pattern for complex multi-step tasks.

Repo: https://github.com/trilwu/secskills

Other community resources

  • Eyadkelleh/awesome-claude-skills-security — curated skill index
  • transilienceai/communitytools — community skills, agents, slash commands
  • dmore/claude-bug-bounty-ai-skill-claude-code — fork of shuvonsec’s
  • travisvn/awesome-claude-skills — general awesome-list
  • VoltAgent/awesome-claude-code-subagents — penetration-tester subagent pattern

Tooling

PortSwigger — Burp Suite + MCP Server extension

Burp Suite Pro/Community is the foundation HTTP intercept tool. Their BApp Store includes an “MCP Server” extension that exposes Burp’s proxy history to Claude Code via the Model Context Protocol.

Burp Suite: https://portswigger.net/burp MCP Server extension: install via Burp’s BApp Store

Anthropic — Claude Code, Skills, MCP

The platform itself.

Claude Code: https://claude.ai/download Skills documentation: https://code.claude.com/docs/en/skills MCP documentation: https://docs.claude.com/en/docs/build-with-claude/mcp

HackerOne API + Bugcrowd VRT

  • HackerOne’s public disclosure program enabled the author to curate the per-class hunt skills from real-world report data.
  • Bugcrowd VRT (Vulnerability Rating Taxonomy) is referenced extensively in bugcrowd-reporting/SKILL.md.

Validation

Built and validated through authorized engagements:

Engagement 1 — Authorized bug-bounty program

Exposed four bug-bounty capability gaps that the author’s contributions directly address:

  1. Hypothesis discipline (validation before drafting) — addressed by triage-validation (vendored)
  2. Per-program reporting tactics — addressed by bugcrowd-reporting (original)
  3. Engagement coordination / scaffolding — addressed by the hunt shell command (original)
  4. Evidence hygiene / redaction — addressed by evidence-hygiene (original)

Engagement 2 — External red-team engagement

Authorized external red-team engagement against an enterprise target. Exposed five additional gaps that bug-bounty defaults made worse:

  1. Conservative defaults retracted real findings → addressed by redteam-mindset (original)
  2. No mid-engagement situational awareness (client patched SQLi in 30 min; external attacker locked 14 accounts mid-test) → addressed by mid-engagement-ir-detection (original)
  3. No enterprise-platform attack chains for M365, on-prem SharePoint, SSL VPN, vCenter, APKs → addressed by m365-entra-attack, okta-attack, hunt-sharepoint, hunt-aspnet, hunt-ntlm-info, vmware-vcenter-attack, enterprise-vpn-attack, apk-redteam-pipeline (all original)
  4. No client-facing deliverable format → addressed by redteam-report-template (original)
  5. No post-credential escalation model → addressed by cloud-iam-deep (original)

Engagement-specific identifiers (target names, domains, account UIDs, IPs, endpoint names, internal app names, employee names, tenant IDs, and any other client-identifying data) have been replaced with abstract placeholders in the shipped versions of all engagement-derived skills. Engagement details are not redistributed.

License notes

  • Original work in this repo: MIT (see LICENSE)
  • Vendored upstream skills retain their original licenses — typically MIT but verify each upstream source above
  • If you’re an upstream author and want attribution adjusted, removed, or expanded, open an issue
This site uses Just the Docs, a documentation theme for Jekyll.