Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 79 600 files from 1 763 authors, of which 61 947 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
Use when design is complete and you need detailed implementation tasks for engineers with zero codebase context - creates comprehensive implementation plans with exact file paths, complete code examples, and verification steps assuming engineer has minimal domain knowledge
Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dependencies, and rust-documentation skills.
Base MCP — gives your AI assistant access to a Base Account via the Base MCP server (mcp.base.org). Wallet, portfolio, sending, swapping, signing, x402 payments, batched contract calls, and transaction history across supported chains.
MapStage live tuner for map/globe projection, satellite, hypsometric relief, 3D terrain, region isolate, CSS color grade, and JSON presets.
Capture a Burp Suite Repeater request/response as a PoC image (targets/<eng>/poc/) by driving the Burp MCP + the Kali GUI. Replays a request in a Repeater tab, sends it, and grabs the request+response panes - a Burp-native PoC (client report / CTF writeup). Use when you want the evidence to come from Burp rather than a curl/terminal card, or whenever you drive a target through Burp and need the images. Pairs with hunt-burp.
Drive Burp Suite over its MCP server as an AI triage + attack layer - review proxy history for signals, replay via Repeater/send, OOB-gate blind bugs with Collaborator, fuzz via Intruder (RoE-safe), then hand off to the matching vuln-class hunt. Wiki-first, FIND schema output.
Drive responsible disclosure of a proven finding to a CVE. Package the report, find the vendor contact, report privately, coordinate a timeline, request the CVE (vendor CNA / GitHub / MITRE), and publish an advisory. Closes the research loop. Triggers - "disclose", "request a cve", "report this to the vendor".
Business-logic flaw hunting - workflow/state bypass, price/quantity tampering, negative/overflow values, coupon/refund abuse, mass assignment, and logic races. The top-paying bug class with no scanner coverage. Wiki-first, FIND schema output.
OFFLINE FALLBACK for the claude-md-management plugin - prefer that plugin when it is installed. Audit and improve CLAUDE.md files - scan for CLAUDE.md files, evaluate quality against templates, output a report, then make targeted updates. Invoke explicitly (/claude-md-improver) when the plugin is unavailable.
Auth bypass and ATO hunting - legacy protocol matrix (XMLRPC, SharePoint /_vti_bin/, EWS, Citrix, etc.), JWT manipulation, password reset poisoning, SAML auth bypass, session fixation. Wiki-first, FIND schema output.
API attack hunting (REST / GraphQL / gRPC) - BOLA/IDOR, BFLA, mass assignment, excessive data exposure, auth/JWT, introspection + batching, rate-limit abuse. OWASP API Top 10. Wiki-first, FIND schema output.
Active Directory attack hunting - enumeration to domain dominance. Spray-safe (lockout gate), AS-REP/Kerberoast, ACL + ADCS (ESC1-16), delegation, DCSync, lateral movement. Wiki-first, FIND schema output.
CI/CD pipeline attack hunting (GitHub Actions focus) - pwn requests (pull_request_target), script injection, self-hosted runner takeover, cache poisoning, OIDC-to-cloud token theft, poisoned pipeline execution. Wiki-first, FIND schema output.
Web cache poisoning + cache deception hunting - unkeyed input poisoning, cache-key analysis, path-confusion deception, header/parameter cloaking. Wiki-first, FIND schema output.
Cloud attack hunting for AWS / Azure / GCP - credential discovery, metadata SSRF, IAM privesc, service enumeration, persistence. Scope + billing aware. Wiki-first, FIND schema output.
> the confirmation gate that separates a real finding from a false positive, enumeration limits, stop conditions, marker discipline, wiki-first query and self-heal, FIND output, Deadends, and wiki distillation. ALWAYS LOADED alongside any hunt skill. Also trigger directly on "is this in scope", "is this a real bug", "how do I confirm this", "should I keep going", "how many IDs should I test", "what severity", "how do I report this", "I got someone else's data". Every hunt-* skill assumes this file; without it they run without their safety and quality layer.
ICS/SCADA/OT exploitation - Modbus (502), S7comm (102), EtherNet/IP (44818), DNP3, OpenPLC, Node-RED SCADA, PLC/HMI/coil/holding-register attacks. Use when a target exposes industrial protocols or the goal is to drive a plant to a dangerous state (over-pressure/over-speed/disable interlock) and read the flag the HMI/CCTV reveals.
Insecure deserialization hunting across Java / .NET / PHP / Python / Ruby / Node. Gadget-chain RCE, OOB-gated blind detection, magic-byte fingerprinting. Wiki-first, FIND schema output.
> IDOR / BOLA hunting - two-account methodology, identifier discovery and UUID leak chaining, the trusted-identifier test, GraphQL node and nested-object IDOR, cross-tenant escalation, write and delete operations. Bounded ID sampling, never range sweeps. Wiki-first, FIND schema output. Trigger on IDOR, BOLA, broken access control, object level authorization, cross-tenant, "read another user's data", "swap the id", or any API path or parameter carrying a numeric ID, UUID, or account identifier.
GraphQL IDOR/auth-bypass, XXE file-read/SSRF (SVG/DOCX/SAML), SSTI detection and RCE. OOB-mandatory for blind XXE. Wiki-first, FIND schema output.
Microsoft 365 / Entra ID attack - tenant discovery, user enumeration via OneDrive differential (2026 verified), AADSTS code reference, Smart Lockout math (hard cap 1-2 attempts/user), ROPC validation, Conditional Access mapping. Wiki-first, FIND schema output.
macOS attack hunting - foothold to root/persistence on a macOS host. TCC/Gatekeeper/SIP bypass, keychain + credential loot, code-signing/entitlements abuse, XPC/dylib/library injection, launch-constraint evasion, MDM/installer abuse. Wiki-first, FIND schema output.
RCE hunting - template injection, YAML/XML deserialization, dependency confusion, Kubernetes surfaces, CVE-specific exploits (Apache CVE-2021-41773, Spring CVE-2022-22963). OOB-mandatory for blind cases. Wiki-first, FIND schema output.
MCP server attack hunting - tool poisoning, indirect prompt injection via tool output, rug-pull updates, cross-tool shadowing, over-permissioned/excessive-agency tools, lethal trifecta. Wiki-first, FIND schema output.
LLM / AI application attack hunting - prompt injection (direct + indirect), excessive agency, insecure output handling, system-prompt + data leakage. OWASP LLM Top 10. Wiki-first, FIND schema output.
Exposed-secrets hunting - .git/ dir + history mining, exposed .env/config files, hardcoded keys in JS bundles + source maps, S3/blob exposure, public-repo secret search, CI/CD leakage. Live-validation mandatory. Wiki-first, FIND schema output.
OAuth and SAML attack hunting - redirect_uri bypass, state CSRF, SAML XSW (XSW1-XSW8), signature stripping, comment injection. Wiki-first, FIND schema output.
HTTP request smuggling / desync hunting - CL.TE, TE.CL, TE.TE, CL.0, and HTTP/2 downgrade. Timing-based detection, differential confirmation, no-blind-claims. Wiki-first, FIND schema output.
SQLi and NoSQLi hunting - error-based, boolean-blind, time-based, UNION, NoSQL operator injection. sqlmap automation after manual confirmation. Wiki-first, FIND schema output.
SSRF hunting - OOB-mandatory methodology. Cloud metadata, blind SSRF via Collaborator/interactsh, redirect-based bypass, headless browser chains. Wiki-first, FIND schema output.
Enterprise SSL VPN attack - vendor fingerprinting, CVE matrix (Cisco, Fortinet, Citrix, Palo Alto, Pulse/Ivanti), default credentials, pre-auth exploit commands. Wiki-first, FIND schema output.
File upload attack hunting - extension/content-type/magic-byte bypass to web-shell RCE, path traversal in filename, SVG/XML XSS, zip slip, and pixel-flood DoS. Wiki-first, FIND schema output.
Local Windows privilege escalation on a STANDALONE / workgroup host, or a local shell on a domain member - foothold to SYSTEM. Token privileges (SeImpersonate/Potato), service misconfig (weak perms / unquoted path / writable binary), registry autologon creds, scheduled-task + writable-script abuse, DLL hijack, AlwaysInstallElevated, UAC bypass, credential loot. For DOMAIN escalation (kerberoast/DCSync/ADCS/BloodHound) use hunt-ad instead. Wiki-first, FIND schema output.
XSS hunting - reflected, stored, DOM-based. Marker discipline to avoid false positives. Blind-XSS beacons for stored contexts. SVG/markdown/redirect vectors. Wiki-first, FIND schema output.
Search, query, and maintain the qmd-indexed wiki - semantic search, keyword search, re-index after adding pages, check index status.
Vulnerability-research loop toward a novel CVE. Target triage -> attack-surface map -> ranked hypotheses -> investigate (RE / fuzz / audit) -> a finding deepens the loop, a dead-end pivots to a new approach. Uses the full wiki + hunt skillset. Scaffolds and persists state under raw/research/<project>/. Triggers - "research", "find a cve", "analyze this binary/library", "audit this code for vulns".
Autonomous bug-bounty campaign driver. Runs a full programme end to end with no operator approvals - the deterministic driver (scripts/campaign.py) owns pass state, generates the killchain board from recon, and prints the exact next action (including which Skill and tool to run) every turn. Use when starting or resuming a bug-bounty engagement, "run the bb workflow", "hunt this program", "9-pass campaign", or when handed a *.scope wildcard to test for TIER1 findings. Single agent, refuter-verified, wiki-first, tool-first.
Health check for the bb/pt/ctf workflow driver subsystem - verifies everything is in place so every machine runs the same. Checks vault-content consistency (scripts present, JSON valid, routing wired, all 69 tool pages carry phase:, the tool index resolves, the hook edits are in place) AND per-machine wiring (the three workflow skills symlinked, hooks registered, imports work), then runs a live init->board->next smoke test. Use when setting up the workflow on a new machine, after a vault sync, when the driver misbehaves, or on "bb-health", "campaign health", "is the workflow set up", "check hooks and scripts", "why is the board not working".
Bring up a REAL, visible, interactive chromium on the Kali VM that the operator logs into (Smart-ID / Mobile-ID / any manual auth or MFA/CAPTCHA), while the agent drives and observes it live through the chrome-devtools MCP (navigate, DOM snapshot, network capture, screenshots, console, evaluate). Use whenever a target needs a MANUAL login the agent cannot complete headlessly, when you need to capture an authenticated session / the real API calls a page makes, or to confirm/screenshot a DOM-XSS. Triggers - "open a browser", "log in manually", "smart-id / mobile-id / national id login", "mfa / 2fa login", "solve the captcha", "drive the browser", "capture the authenticated session / network".
Wiki-first "what do I use" lookup - pick the automated TOOL (wiki/tools/), then the PAYLOAD/technique (wiki/payloads/ + wiki/cheatsheets/), for a surface/service/vuln-class BEFORE hand-rolling or working from memory. Use for "tool for <service>", "automated tools for web/<service>", "what should I run on <surface>", "which tool for <X>", "payloads for <X>", "payload arsenal", "cheatsheet for <X>", "how do I exploit <tech/class>", "exploit/attack chain for <X>", "arsenal", any SSRF/XSS/SQLi/SSTI/LFI/JWT/XXE/IDOR/NoSQL/deserialization/CSRF/CORS/CRLF/GraphQL/smuggling/web-cache/OAuth/SAML/MFA/crypto/LDAP/XPath/WebAuthn/file-upload/IMDS/prompt-injection/Modbus ask, plus "privesc arsenal", "CVE arsenal", "default creds", "nuclei templates", "sqlmap/hydra/nmap/bloodhound", "password attacks".
Show per-asset vuln-class coverage gaps for the active engagement so nothing in scope is skipped. Use when asked "coverage", "what haven't we tested", "test gaps", "are we thorough", or before calling an engagement done.
Boot-to-root methodology for a full machine (THM/HTB/PG/CTF box, "get user.txt+root.txt", "root the box", "foothold to root"). Enforces basic-tool recon (nmap, nc, ffuf, nuclei, dig) before anything custom, wiki-first lookups, and ALWAYS pspy + linpeas/winpeas for privesc. Use when handed a box/IP to own end-to-end.
CTF challenge router - fingerprint a challenge (file type / prompt / artifacts) into its category (pwn, rev, crypto, forensics, stego, web, osint, hash) and route to the matching wiki page, tools, and first moves. Wiki-first.
Autonomous CTF / boot-to-root campaign driver. Runs a box end to end with no operator approvals - the deterministic driver (scripts/campaign.py) owns pass state, generates the killchain board from recon, and prints the exact next action (Skill + tool) every turn. Use when handed a box/IP to own end to end, "run the ctf workflow", "root this box", "foothold to root". Single agent, wiki-first, tool-first. OSINT is OFF unless you invoke with an osint argument. Delegates box recipes to ctf-box.
Autonomous sub-agent hand-off for a fiddly, fully-specified exploit-compile / escalation RUN - the main agent stays on strategy and the board while a cheap sub-agent runs an exact copy-paste checklist behind a false-root/hostname guardrail. Use for "delegate", "offload", "hand this to a sub-agent", "spin a haiku", or the moment a foothold plus a working escalation vector is identified. Main agent dispatches, waits (no parallel duplicate), integrates the result.
Evidence hygiene before any FIND moves to Completed or enters a report. Cookie redaction, PII black-bar, HAR sanitization, screenshot metadata strip. Run after /triage passes and before final report assembly.
Synthesize raw recon/test output into engagement state. Reads everything dropped in targets/<active>/ingest/, extracts hosts/assets/creds/paths, merges into state.md/loot.md/Killchain.md, logs it, archives the raw files. Works for pentest, bugbounty, and ctf. Use when asked to "ingest", "synthesize findings", "process recon", or after dropping tool output in the ingest folder.
Adaptive, targeted web fuzzing - deterministic wordlist selection (wl-pick.sh) plus judgment. Picks the right SecLists list per surface (content/vhost/api/params/artifacts) smallest-first, calibrates filters against soft-404s, recurses, escalates T0 harness -> T1 seclists -> T2 cewl -> T3 app-specific on signal, pivots to hidden-param fuzzing, and detects/handles WAF/Cloudflare/throttle (backoff, origin-bypass, or hard STOP on the DoS tell). Engagement-type aware (ctf loud, pt calibrated, bb stealth). Use for "fuzz", "content discovery", "directory brute", "vhost fuzz", "find hidden params", "which wordlist", "gobuster/ffuf/feroxbuster/cewl/arjun".
Post-engagement knowledge harvest AND harness retrospective - after a box/bugbounty/pentest/CTF is completed, first diff how the engagement was EXECUTED against the skills/hooks that governed it (what discipline was skipped) and improve the harness, then sweep the whole engagement for GENERIC reusable knowledge NOT already in wiki/ and land it via the leak-gated stage->promote pipeline. Use at close-out or when asked to "extract learnings", "what did we learn", "harvest lessons into wiki", "distill this engagement", "post-mortem into the wiki", "what did we do wrong", "improve the harness from this box".
Drive msfconsole across the workflow - DB-backed recon (db_nmap, auxiliary scanners), version->exploit search/check/run, multi/handler reverse shells (meterpreter-first, plain shell_reverse_tcp backup for Windows/EDR), sessions + local_exploit_suggester + post modules, and autoroute/portfwd/socks pivoting. Points to the metasploit cheatsheet for syntax. Use for "metasploit", "msfconsole", "msfvenom", "meterpreter", "multi/handler", or driving an exploit/reverse-shell through msf.
N-day / patch-diff workflow - given a CVE/advisory or a suspicious patch, diff pre- vs post-patch to locate the fixed bug, build a PoC for the unpatched version, and run variant analysis for a fresh bug. Triggers - "n-day", "patch diff", "diff the patch", "bindiff".
On-demand senior red-team lead advisor. Call at a decision point or obstacle to get wiki-grounded direction instead of hammering blindly. Dispatches a fresh RTL subagent that reads the engagement state + evidence + wiki and returns ranked directions with an explicit STOP. Use for "redteamlead", "RTL", "I'm stuck", "where do I go", "what next", "which vector", "should I keep hammering this".
Ranked next offensive moves from engagement state. Reads state/loot/Killchain, runs the deterministic analyzer, elaborates the top move. Use when asked "what next", "where to focus", "prioritize", or at the start of an engagement session.
Autonomous pentest campaign driver. Runs a scoped engagement end to end with no operator approvals - the deterministic driver (scripts/campaign.py) owns pass state, generates the killchain board from recon, and prints the exact next action (Skill + tool) every turn. Use when starting or resuming a pentest, "run the pt workflow", "work this CIDR/domain", or when handed a client SoW/scope to reach a stated objective or domain admin. Single agent, refuter-verified, wiki-first, tool-first. Deliverable is a client report.
Capture web-page / PoC screenshots into the engagement evidence (targets/<eng>/poc/) and embed them in walkthrough.md. Live pages + authenticated/exploited states (post-login dashboard, the flag page, an SSTI/cmdi render). Runs chromium on the Kali tooling host (VPN path to targets); hands off to the evidence skill for redaction before a real report. Use when building PoC evidence or asked to "take a screenshot / capture the PoC".
Finding validation gate - 7-Question triage adapted for FIND schema. Run before moving any FIND from Research to Completed. One NO = keep in Research. All YES = move to Completed, then run /evidence.
Assemble a report-ready walkthrough.md for a SOLVED engagement - populate the Evidence gallery from the poc/ images captured during the engagement, and draft the step-by-step reproduction from state/loot/log (Killchain.md for pentest/bugbounty, state.md's Chain/Status for ctf) without fabricating. Use when asked to "write the walkthrough", "assemble the walkthrough", "close out the box/engagement", or at close-out once an engagement is marked SOLVED.
Ingest a CVE writeup, blog post, advisory, or GitHub repo into the wiki - fetch, dedup via sources:, update the right technique/tool page(s), re-index. Generic knowledge only; never client data.
Fast PARALLEL wiki lookup engine over wiki/techniques + wiki/payloads + wiki/tools + wiki/cheatsheets for a surface/service/vuln-class. Two modes - quick (one qmd search, cheap, fire constantly) and deep (4 parallel subagents, one per area, merged ready-to-use arsenal card, cached). This is the fast path arsenal that `arsenal` delegates to; the hunt-* skills each inline their own qmd_query and can hand off here for a parallel lookup. Use for "what do I use against <surface>", "arsenal for <X>", "deep/full arsenal", "tool + payload + technique + cheatsheet for <X>", "fast wiki lookup", "parallel wiki search", any "how do I attack/exploit <service|vuln-class>" where you want the documented tooling + payloads before hand-rolling.
External recon and OSINT pipeline - subdomain enum, live host discovery, URL crawl, JS analysis, nuclei scan. Outputs to Attack-surface.md and scope/. Queries wiki before each phase. Use when starting recon on any target.
Claude Skills — questions
Answers built from the skills we actually parsed.