AGENT SKILL

solve-challenge

Solves CTF challenges by performing first-pass triage, identifying the dominant category, and routing execution to the right specialized ctf-* skill. Use when the user gives you a challenge bundle, a remote service, a suspicious file, or only a vague challenge description and you must determine where to start. Do not use it when the category is already clear and a specialized skill can be invoked directly; this is the dispatcher and recon entrypoint, not the deepest reference for category-specific techniques.

License 许可
MIT
最近更新
2026-11-15
Stars
★ 81
Forks
17

01 前言 Front Matter

name
solve-challenge
license
MIT
metadata
{ "author": "Lukasz Jagiello", "argument-hint": "[category] [challenge-file-or-url]", "user-invocable": "true" }
description
Solves CTF challenges by performing first-pass triage, identifying the dominant category, and routing execution to the right specialized ctf-* skill. Use when the user gives you a challenge bundle, a remote service, a suspicious file, or only a vague challenge description and you must determine where to start. Do not use it when the category is already clear and a specialized skill can be invoked directly; this is the dispatcher and recon entrypoint, not the deepest reference for category-specific techniques.
allowed-tools
Bash Read Write Edit Glob Grep Task WebFetch WebSearch Skill
compatibility
Requires filesystem-based agent (Claude Code or similar) with bash, Python 3, and internet access. Orchestrates other ctf-* skills.

02 资源文件 4

03 SKILL.md 正文

CTF Challenge Solver

You're a skilled CTF player. Your goal is to solve the challenge and find the flag.

Environment Setup

Two setup strategies depending on your workflow:

Pre-install (recommended before competitions)

Use the central installer entrypoint:

bash scripts/install_ctf_tools.sh all

Run a narrower mode when you only want one tool group:

bash scripts/install_ctf_tools.sh python
bash scripts/install_ctf_tools.sh apt
bash scripts/install_ctf_tools.sh brew
bash scripts/install_ctf_tools.sh gems
bash scripts/install_ctf_tools.sh go
bash scripts/install_ctf_tools.sh manual

The full package lists now live in scripts/install_ctf_tools.sh.

On-demand (during challenges)

Each category skill's SKILL.md has a Prerequisites section listing only the tools needed for that category. Install as you go.

Workflow

Before selecting or escalating a model, read difficulty-routing.md. Use its evidence-based scoring, stop-loss rules, and tier gates throughout the workflow.

Step 0: CTFd Platform Detection

If the CTF platform URL is known, check if it runs CTFd and switch to API-driven navigation:

# Detect CTFd (look for /api/v1/ and /themes/core/)
curl -s "$CTF_URL/api/v1/" | head -5
curl -s "$CTF_URL" | grep -oE '/themes/core/'

If CTFd is detected, ask the user for their API token (generated from CTFd Settings > Access Tokens). The token is not provided by default — the user must create one in the CTFd web UI first. Once provided, set the environment variables and proceed via API:

export CTF_URL="https://ctf.example.com"
export CTF_TOKEN="ctfd_..."  # Ask user for this

Invoke /ctf-misc and load its ctfd-navigation.md for the full API reference and Python client class.

CTF Platform Interaction via Playwright MCP

When the platform is not CTFd, has no usable API, or gates challenges behind a JS-rendered UI / login, drive the platform in a real browser with Playwright MCP (https://github.com/microsoft/playwright-mcp). This lets the agent read challenge text, download attachments, keep a login session, and submit flags directly through the site.

browser_navigate "https://ctf.example.com/challenges"
browser_snapshot                 # list challenges, points, categories (rendered)
browser_click <challenge ref>    # open a challenge card
browser_snapshot                 # read description, hints, attachment links
browser_click <attachment ref>   # download the challenge file (saved by the browser)
# ... solve the challenge ...
browser_type <flag input ref> "flag{...}"
browser_click <submit ref>
browser_find text="Correct"      # cheaper than a full snapshot to check the verdict

browser_find (text or regex over the accessibility tree) is the right tool whenever you only need to confirm one value — reserve full browser_snapshot for when you need element refs to click or type into.

Login sessions (cookies/localStorage) persist across tool calls, so authenticate once and every later navigation is authorized. For general browser-automation tooling see ../ctf-web/tools-mcp.md.

Prefer a native API when one exists (CTFd token flow above is faster and more reliable); use Playwright MCP for platforms without an API or when the challenge list only renders client-side.

Step 1: Recon

  1. Explore files -- List the challenge directory, run file * on everything
  2. Triage binaries -- strings, xxd | head, binwalk, checksec on binaries
  3. Fetch links -- If the challenge mentions URLs, fetch them FIRST for context
  4. Connect -- Try remote services (nc) to understand what they expect
  5. Read hints -- Challenge descriptions, filenames, and comments often contain clues

Step 2: Categorize

Determine the primary category, then invoke the matching skill.

By file type:

  • .pcap, .pcapng, .evtx, .raw, .dd, .E01 -> forensics
  • .elf, .exe, .so, .dll, binary with no extension -> reverse or pwn (check if remote service provided -- if yes, likely pwn)
  • .py, .sage, .txt with numbers -> crypto
  • .apk, .wasm, .pyc -> reverse
  • firmware/image (firmware.bin, .img, vendor update), .sr/.vcd logic capture, CAN log -> hardware
  • .safetensors, .pt, .pth, .bin, .onnx -> ai-ml
  • Web URL or source code with HTML/JS/PHP/templates -> web
  • Images, audio, PDFs with no obvious content -> forensics (steganography)

By challenge description keywords:

  • "buffer overflow", "ROP", "shellcode", "libc", "heap" -> pwn
  • "RSA", "AES", "cipher", "encrypt", "prime", "modulus", "lattice", "LWE", "GCM" -> crypto
  • "XSS", "SQL", "injection", "cookie", "JWT", "SSRF" -> web
  • "disk image", "memory dump", "packet capture", "registry", "power trace", "side-channel", "spectrogram", "audio tracks", "MKV" -> forensics
  • "find", "locate", "identify", "who", "where" -> osint
  • "obfuscated", "packed", "C2", "malware", "beacon" -> malware
  • "prompt injection", "LoRA", "adversarial", "model weights", "embedding" -> ai-ml
  • "jail", "sandbox", "escape", "restricted shell", "no quotes", "eval" -> jail
  • "metadata", "IAM", "S3", "kubectl", "container escape", "docker.sock" -> cloud
  • "firmware", "UART", "JTAG", "SPI flash", "logic analyzer", "power trace", "CAN bus" -> hardware
  • "encoding", "signal", "game", "Nim", "commitment", "Gray code" -> misc

By service behavior:

  • Port with interactive prompt, crash on long input -> pwn
  • HTTP service -> web
  • netcat with math/crypto puzzles -> crypto
  • netcat with restricted shell or eval -> misc (jail)

Step 3: Invoke the Category Skill

Once you identify the category, invoke the matching skill to get specialized techniques:

Category Invoke When to Use
Web /ctf-web XSS, SQLi, SSTI, SSRF, JWT, file uploads, prototype pollution
Pwn /ctf-pwn Buffer overflow, format string, heap, ROP, sandbox escape
Crypto /ctf-crypto RSA, AES, ECC, PRNG, ZKP, classical ciphers
Reverse /ctf-reverse Binary analysis, game clients, VMs, obfuscated code
Forensics /ctf-forensics Disk images, memory dumps, event logs, stego, network captures
OSINT /ctf-osint Social media, geolocation, DNS, public records
Malware /ctf-malware Obfuscated scripts, C2 traffic, PE/.NET analysis
AI/ML /ctf-ai-ml Model weights (.safetensors, .pt, .pth, .bin, .onnx), prompt injection, LoRA adapters, adversarial examples
Jail /ctf-jail Python/bash jails, restricted eval, sandbox escapes, disable_functions
Cloud /ctf-cloud Cloud metadata, IAM, S3, Kubernetes, container escape, CI/CD
Hardware /ctf-hardware Firmware, UART/JTAG/SPI, logic capture, CAN, side channel
Misc /ctf-misc Encodings, RF/SDR demod, esoteric languages, constraint solving, games

You can also invoke /ctf-<category> to load the full skill instructions with detailed techniques.

Step 4: Pivot When Stuck

If your first approach doesn't work:

  1. Re-examine assumptions -- Is this really the category you think? A "web" challenge might need crypto for JWT forgery. A "forensics" PCAP might contain a pwn exploit to replay.
  2. Try a different category skill -- Many challenges span multiple categories. Invoke a second skill for the cross-cutting technique.
  3. Look for what you missed -- Hidden files, alternate ports, response headers, comments in source, metadata in images.
  4. Simplify -- If an exploit is too complex, check if there's a simpler path (default creds, known CVE, logic bug).
  5. Check edge cases -- Off-by-one, race conditions, integer overflow, encoding mismatches.

Common multi-category patterns:

  • Forensics + Crypto: encrypted data in PCAP/disk image, need crypto to decrypt
  • Web + Reverse: WASM or obfuscated JS in web challenge
  • Web + Crypto: JWT forgery, custom MAC/signature schemes
  • Reverse + Pwn: reverse the binary first, then exploit the vulnerability
  • Forensics + OSINT: recover data from dump, then trace it via public sources
  • Misc + Crypto: jail escape requires building crypto primitives under constraints
  • OSINT + Stego: social media posts with unicode homoglyph steganography (Cyrillic lookalikes encode bits)
  • Web + Forensics: paywall bypass (curl reveals content hidden by CSS overlays)
  • Misc + Crypto + Game Theory: multi-phase interactive challenges with AES decryption → HMAC commitment → combinatorial game solving (GF(256) Nim)
  • Crypto + Geometry + Lattice: multi-layer challenges progressing from spatial reconstruction → subspace recovery → LWE solving → AES-GCM decryption
  • Forensics + Signal Processing: power traces / side-channel analysis requiring statistical analysis of measurement data
  • Forensics + Network + Encoding: timing-based encoding in PCAP (inter-packet intervals encode binary data)

Step 5: Generate Write-up

After solving the challenge, invoke /ctf-writeup to generate a standardized submission-style writeup — concise, reproducible, and ready for competition organizers or teammates to validate.

Flag Formats

Flags vary by CTF. Common formats:

  • flag{...}, FLAG{...}, CTF{...}, TEAM{...}
  • Custom prefixes: check the challenge description or CTF rules for the format (e.g., ENO{...}, HTB{...}, picoCTF{...})
  • Sometimes just a plaintext string with no wrapper

Validation rule (important):

  • If you find multiple flag-like strings, treat them as candidates and validate before finalizing.
  • Prefer the token tied to the intended artifact/workflow (not random metadata noise or obvious decoys).
  • Do a corpus-wide uniqueness check and include the source file/path when reporting.
# Search for common flag patterns in files
grep -rniE '(flag|ctf|eno|htb|pico)\{' .
# Search in binary/memory output
strings output.bin | grep -iE '\{.*\}'

Quick Reference

# Recon
file *                                    # Identify file types
strings binary | grep -i flag             # Quick string search
xxd binary | head -20                     # Hex dump header
binwalk -e firmware.bin                   # Extract embedded files
checksec --file=binary                    # Check binary protections

# Connect
nc host port                              # Connect to challenge
echo -e "answer1\nanswer2" | nc host port # Scripted input
curl -v http://host:port/                 # HTTP recon

# Python exploit template
python3 -c "
from pwn import *
r = remote('host', port)
r.interactive()
"

Challenge

$ARGUMENTS

04 同仓库更多 Skills

jiayaoqijia
jiayaoqijia/cryptoskill

troves

Deposit and withdraw from Troves DeFi yield vault strategies on Starknet. View available strategies, check positions, and manage vault deposits. Use this skill when the user mentions Troves, vault, yield vault, DeFi vault, strategy, yield farming, vault deposit, vault withdraw, earning yield through vaults, passive yield strategies, or wants to deposit into or withdraw from a yield strategy — even if they don't say "Troves" explicitly.

★ 81
jiayaoqijia
jiayaoqijia/cryptoskill

trade

Swap tokens on Starknet via Fibrous DEX aggregator. Supports per-provider selection (avnu, ekubo, or auto to race all). Use this skill when the user wants to swap, exchange, trade, convert, buy, or sell one token for another on Starknet. Also use when the user asks about token prices, exchange rates, DEX routing, best swap route, or wants to check how much they would receive for a given amount — even if they don't explicitly say "swap".

★ 81
jiayaoqijia
jiayaoqijia/cryptoskill

staking

Stake, unstake, claim, and compound rewards for STRK, WBTC, tBTC, SolvBTC, and LBTC on Starknet. View validators, pools, and staking positions. Use this skill when the user mentions staking, delegating, validators, staking pools, rewards, compounding, unstaking, earning yield, passive income on crypto, or wants to earn on their STRK or BTC tokens — even if they don't say "stake" explicitly.

★ 81
jiayaoqijia
jiayaoqijia/cryptoskill

send

Send STRK, ETH, or any ERC-20 token to a Starknet address. Supports simulation (dry-run) before execution. Use this skill when the user wants to send, transfer, pay, tip, or move tokens to another wallet address on Starknet. Also trigger when the user mentions sending funds to someone, making a payment, transferring assets, or moving crypto to a specific address — even if they just say "pay 10 STRK to 0x...".

★ 81
jiayaoqijia
jiayaoqijia/cryptoskill

portfolio

View a comprehensive DeFi portfolio dashboard and optimize portfolio allocation via automated rebalancing. Use this skill when the user wants an overview, summary, dashboard, total value, net worth, portfolio rebalancing, target allocation, or wants to see all their DeFi positions across Starknet at once. Also trigger when the user asks "what do I have", "show me everything", "how much am I worth", "my positions", "my investments", "rebalance my portfolio", "I want 50% ETH", "optimize my holdings", or any request for a holistic view of their Starknet holdings — even if they don't explicitly say "portfolio".

★ 81
jiayaoqijia
jiayaoqijia/cryptoskill

multi-swap

Execute multiple token swaps in a single Starknet transaction via Fibrous (default). Supports per-provider selection (avnu, ekubo, or auto to race all). Supports 2-3 swap pairs bundled into one multicall. Use this skill when the user wants to swap multiple token pairs at once, do batch swaps, execute several trades simultaneously, perform parallel swaps, or do bulk trading in one transaction. Also trigger when the user mentions "two swaps", "three swaps", "swap X and Y at the same time", or wants to convert multiple tokens in a single call.

★ 81