andresriancho

from typing import Optional, List
from llmlingua import PromptCompressor

MODEL_CONFIG = {}
MODEL = "microsoft/llmlingua-2-xlm-roberta-large-meetingbank"
RATE = 0.33

class LLMLinguaSegment:
    """Class representing a single <llmlingua> segment, encapsulating its content, rate, and compress flag."""
    

andresriancho

import boto3
import botocore

def log_request(request, **kwargs):
    """Logs all HTTP requests made by boto3 before they are sent."""
    print(f"🔍 Intercepted HTTP request:")
    print(f"  - Method: {request.method}")
    print(f"  - URL: {request.url}")
    print(f"  - Headers: {request.headers}")
    print(f"  - Body: {request.body}\n")

andresriancho

SELECT eventtime,
         eventname,
         requestparameters,
         awsregion,
         eventsource,
         resources
FROM cloudtrail_662651605507
WHERE year = '2019'
        AND month IN ('7', '8', '9', '10', '11')
        AND eventsource = 's3.amazonaws.com'

andresriancho

SELECT eventtime,
         eventname,
         requestparameters,
         awsregion,
         eventsource,
         resources
FROM cloudtrail_662651605507
WHERE year = '2019'
        AND month IN ('7', '8', '9', '10', '11')
        AND eventsource = 'lambda.amazonaws.com'

andresriancho

GuardDuty events sent via CloudWatch rules and SNS emails are ugly.

It is possible to improve them using inputTransformer.

guardduty-event-target.tf shows an example usage.

andresriancho

import boto3

session = boto3.Session(profile_name='ariancho')
s3_client = session.client('s3')

display_name = s3_client.list_buckets()['Owner']['DisplayName']
print(display_name)

for bucket in s3_client.list_buckets()['Buckets']:
    print(s3_client.get_bucket_acl(Bucket=bucket['Name'])['Owner']['DisplayName'])

andresriancho

import os
import sys
import time
import random
import subprocess

FNULL = open(os.devnull, 'w')

env = {'HTTPS_PROXY': 'http://localhost:8080/'}
cmd = ('aws --region us-east-1 --no-verify-ssl --profile=andres-root s3api delete-object'

andresriancho

import timeit
import lz4.frame
import lzf
import zlib
#import snappy
import os
from timeit import Timer

DATA = open("test.py", "rb").read()
DLEN = len(DATA)

andresriancho

User story

As an application security expert I use Atom to read source code. I would like to be able to have a plugin with features that will help me identify vulnerabilities.

Ideas

• Add annotations to the source code: Select a few lines of code, right click, "Add annotation", text box appears, user types comments on the source code, clicks "Save". All annotations can be seen in a tab. If the code has annotations then it is highlighted differently (change background color)

• The plugin implements methods for searching for XSS, SQL injection, etc. The plugin adds a menu

andresriancho

Research areas

How does Oraclize work? How do they use crypto to guarantee that they are providing unmodified results for the HTTP queries?

Blogs

• http://martin.swende.se/
• https://pdaian.com/blog/solidity-anti-patterns-fun-with-inheritance-dag-abuse/
• https://blog.trailofbits.com/2017/11/06/hands-on-the-ethernaut-ctf/