SkillDevelopment & Engineering
Trace a request through a codebase
Follow one request from entry point to storage or external boundary with concrete file references.
✳Claude Code
By Dave AIDave AI original
MIT · FreeView skill
GDB debugging assistant for AI agents - analyze core dumps, debug live processes, investigate crashes and deadlocks with source code correlation
---
name: gdb-cli
description: "GDB debugging assistant for AI agents - analyze core dumps, debug live processes, investigate crashes and deadlocks with source code correlation"
category: development
risk: critical
source: community
date_added: "2026-03-22"
author: Cerdore
tags:
- debugging
- gdb
- core-dump
- crash-analysis
- c++
- c
tools:
- claude-code
- cursor
- gemini-cli
- codex-cli
- antigravity
---
# GDB Debugging Assistant
## Overview
A GDB debugging skill designed for AI agents. Combines **source code analysis** with **runtime state inspection** using gdb-cli to provide intelligent debugging assistance for C/C++ programs.
## When to Use This Skill
- Analyze core dumps or crash dumps
- Debug running processes with GDB attach
- Investigate crashes, deadlocks, or memory issues
- Get intelligent debugging assistance with source code context
- Debug multi-threaded applications
## Do Not Use This Skill When
- The task is unrelated to C/C++ debugging
- The user needs general-purpose assistance without debugging
- No GDB is available (GDB 9.0+ with Python support required)
## Prerequisites
```bash
# Install gdb-cli
pip install gdb-cli
# Or from GitHub
pip install git+https://github.com/Cerdore/gdb-cli.git
# Verify GDB has Python support
gdb -nx -q -batch -ex "python print('OK')"
```
**Requirements:**
- Python 3.6.8+
- GDB 9.0+ with Python support enabled
- Linux OS
Review/pin gdb-cli and dependencies before installation. For untrusted binaries/cores, disable GDB auto-loading and initialization before loading, use patched tools and resource isolation, and never run the inferior or evaluate function calls just to inspect data. A live attach pauses/changes process timing and requires authorization for that exact PID/service. Backtraces/locals/memory can contain secrets; collect only necessary redacted evidence and retain core files privately. Do not change ptrace/system policy or use sudo as an automatic workaround; verify stop detaches/resumes the intended target.
## How It Works
### Step 1: Initialize Debug Session
**For core dump analysis:**
```bash
gdb-cli load --binary <binary_path> --core <core_path> [--gdb-path <gdb_path>]
```
**For live process debugging:**
```bash
gdb-cli attach --pid <pid> [--binary <binary_path>]
```
**Output:** A session_id like `"session_id": "a1b2c3"`. Store this for subsequent commands.
### Step 2: Gather Initial Information
```bash
SESSION="<session_id>"
# List all threads
gdb-cli threads -s $SESSION
# Get backtrace (with local variables)
gdb-cli bt -s $SESSION --full
# Get registers
gdb-cli registers -s $SESSION
```
### Step 3: Correlate Source Code (CRITICAL)
For each frame in the backtrace:
1. **Extract frame info**: `{file}:{line} in {function}`
2. **Read source context**: Get ±20 lines around the crash point
3. **Get local variables**: `gdb-cli locals-cmd -s $SESSION --frame <N>`
4. **Analyze**: Correlate code logic with variable values
**Example correlation:**
```
Frame #0: process_data() at src/worker.c:87
Source code shows:
85: Node* node = get_node(id);
86: if (node == NULL) return;
87: node->data = value; <- Crash here
Variables show:
node = 0x0 (NULL)
Analysis: This evidence is inconsistent with the apparent source path. Verify the exact
binary/debug-symbol/source build, optimization and frame accuracy; investigate races
or corruption rather than declaring that a valid NULL check failed.
```
### Step 4: Deep Investigation
```bash
# Examine variables
gdb-cli eval-cmd -s $SESSION "variable_name"
gdb-cli eval-cmd -s $SESSION "ptr->field"
gdb-cli ptype -s $SESSION "struct_name"
# Memory inspection
gdb-cli memory -s $SESSION "0x7fffffffe000" --size 64
# Disassembly
gdb-cli disasm -s $SESSION --count 20
# Check all threads (for deadlock analysis)
gdb-cli thread-apply -s $SESSION bt --all
# View shared libraries
gdb-cli sharedlibs -s $SESSION
```
### Step 5: Session Management
```bash
# List active sessions
gdb-cli sessions
# Check session status
gdb-cli status -s $SESSION
# Stop session (cleanup)
gdb-cli stop -s $SESSION
```
## Common Debugging Patterns
### Pattern: Null Pointer Dereference
**Indicators:**
- Crash on memory access instruction
- Pointer variable is 0x0
**Investigation:**
```bash
gdb-cli registers -s $SESSION # Check RIP
gdb-cli eval-cmd -s $SESSION "ptr" # Check pointer value
```
### Pattern: Deadlock
**Indicators:**
- Multiple threads stuck in lock functions
- `pthread_mutex_lock` in backtrace
**Investigation:**
```bash
gdb-cli thread-apply -s $SESSION bt --all
# Look for circular wait patterns
```
### Pattern: Memory Corruption
**Indicators:**
- Crash in malloc/free
- Garbage values in variables
**Investigation:**
```bash
gdb-cli memory -s $SESSION "&variable" --size 128
gdb-cli registers -s $SESSION
```
## Examples
### Example 1: Core Dump Analysis
```bash
# Load core dump
gdb-cli load --binary ./myapp --core /tmp/core.1234
# Get crash location
gdb-cli bt -s a1b2c3 --full
# Examine crash frame
gdb-cli locals-cmd -s a1b2c3 --frame 0
```
### Example 2: Live Process Debugging
```bash
# Attach to stuck server
gdb-cli attach --pid 12345
# Check all threads
gdb-cli threads -s b2c3d4
# Get all backtraces
gdb-cli thread-apply -s b2c3d4 bt --all
```
## Best Practices
- Always read source code before drawing conclusions from variable values
- Use `--range` for pagination on large thread counts or deep backtraces
- Use `ptype` to understand complex data structures before examining values
- Check all threads for multi-threaded issues
- Cross-reference types with source code definitions
## Security & Safety Notes
- This skill requires GDB access to processes and core dumps
- Attaching to processes may require appropriate permissions (sudo, ptrace_scope)
- Core dumps may contain sensitive data - handle with care
- Only debug processes you have authorization to analyze
## Related Skills
- `@systematic-debugging` - General debugging methodology
- `@test-driven-development` - Write tests before implementation
## Links
- **Repository**: https://github.com/Cerdore/gdb-cli
- **PyPI**: https://pypi.org/project/gdb-cli/
- **Documentation**: https://github.com/Cerdore/gdb-cli#readme
## Limitations
- Use this skill only when the task clearly matches the scope described above.
- Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
- Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.
---
name: gdb-cli
description: "GDB debugging assistant for AI agents - analyze core dumps, debug live processes, investigate crashes and deadlocks with source code correlation"
category: development
risk: critical
source: community
date_added: "2026-03-22"
author: Cerdore
tags:
- debugging
- gdb
- core-dump
- crash-analysis
- c++
- c
tools:
- claude-code
- cursor
- gemini-cli
- codex-cli
- antigravity
---
# GDB Debugging Assistant
## Overview
A GDB debugging skill designed for AI agents. Combines **source code analysis** with **runtime state inspection** using gdb-cli to provide intelligent debugging assistance for C/C++ programs.
## When to Use This Skill
- Analyze core dumps or crash dumps
- Debug running processes with GDB attach
- Investigate crashes, deadlocks, or memory issues
- Get intelligent debugging assistance with source code context
- Debug multi-threaded applications
## Do Not Use This Skill When
- The task is unrelated to C/C++ debugging
- The user needs general-purpose assistance without debugging
- No GDB is available (GDB 9.0+ with Python support required)
## Prerequisites
```bash
# Install gdb-cli
pip install gdb-cli
# Or from GitHub
pip install git+https://github.com/Cerdore/gdb-cli.git
# Verify GDB has Python support
gdb -nx -q -batch -ex "python print('OK')"
```
**Requirements:**
- Python 3.6.8+
- GDB 9.0+ with Python support enabled
- Linux OS
Review/pin gdb-cli and dependencies before installation. For untrusted binaries/cores, disable GDB auto-loading and initialization before loading, use patched tools and resource isolation, and never run the inferior or evaluate function calls just to inspect data. A live attach pauses/changes process timing and requires authorization for that exact PID/service. Backtraces/locals/memory can contain secrets; collect only necessary redacted evidence and retain core files privately. Do not change ptrace/system policy or use sudo as an automatic workaround; verify stop detaches/resumes the intended target.
## How It Works
### Step 1: Initialize Debug Session
**For core dump analysis:**
```bash
gdb-cli load --binary <binary_path> --core <core_path> [--gdb-path <gdb_path>]
```
**For live process debugging:**
```bash
gdb-cli attach --pid <pid> [--binary <binary_path>]
```
**Output:** A session_id like `"session_id": "a1b2c3"`. Store this for subsequent commands.
### Step 2: Gather Initial Information
```bash
SESSION="<session_id>"
# List all threads
gdb-cli threads -s $SESSION
# Get backtrace (with local variables)
gdb-cli bt -s $SESSION --full
# Get registers
gdb-cli registers -s $SESSION
```
### Step 3: Correlate Source Code (CRITICAL)
For each frame in the backtrace:
1. **Extract frame info**: `{file}:{line} in {function}`
2. **Read source context**: Get ±20 lines around the crash point
3. **Get local variables**: `gdb-cli locals-cmd -s $SESSION --frame <N>`
4. **Analyze**: Correlate code logic with variable values
**Example correlation:**
```
Frame #0: process_data() at src/worker.c:87
Source code shows:
85: Node* node = get_node(id);
86: if (node == NULL) return;
87: node->data = value; <- Crash here
Variables show:
node = 0x0 (NULL)
Analysis: This evidence is inconsistent with the apparent source path. Verify the exact
binary/debug-symbol/source build, optimization and frame accuracy; investigate races
or corruption rather than declaring that a valid NULL check failed.
```
### Step 4: Deep Investigation
```bash
# Examine variables
gdb-cli eval-cmd -s $SESSION "variable_name"
gdb-cli eval-cmd -s $SESSION "ptr->field"
gdb-cli ptype -s $SESSION "struct_name"
# Memory inspection
gdb-cli memory -s $SESSION "0x7fffffffe000" --size 64
# Disassembly
gdb-cli disasm -s $SESSION --count 20
# Check all threads (for deadlock analysis)
gdb-cli thread-apply -s $SESSION bt --all
# View shared libraries
gdb-cli sharedlibs -s $SESSION
```
### Step 5: Session Management
```bash
# List active sessions
gdb-cli sessions
# Check session status
gdb-cli status -s $SESSION
# Stop session (cleanup)
gdb-cli stop -s $SESSION
```
## Common Debugging Patterns
### Pattern: Null Pointer Dereference
**Indicators:**
- Crash on memory access instruction
- Pointer variable is 0x0
**Investigation:**
```bash
gdb-cli registers -s $SESSION # Check RIP
gdb-cli eval-cmd -s $SESSION "ptr" # Check pointer value
```
### Pattern: Deadlock
**Indicators:**
- Multiple threads stuck in lock functions
- `pthread_mutex_lock` in backtrace
**Investigation:**
```bash
gdb-cli thread-apply -s $SESSION bt --all
# Look for circular wait patterns
```
### Pattern: Memory Corruption
**Indicators:**
- Crash in malloc/free
- Garbage values in variables
**Investigation:**
```bash
gdb-cli memory -s $SESSION "&variable" --size 128
gdb-cli registers -s $SESSION
```
## Examples
### Example 1: Core Dump Analysis
```bash
# Load core dump
gdb-cli load --binary ./myapp --core /tmp/core.1234
# Get crash location
gdb-cli bt -s a1b2c3 --full
# Examine crash frame
gdb-cli locals-cmd -s a1b2c3 --frame 0
```
### Example 2: Live Process Debugging
```bash
# Attach to stuck server
gdb-cli attach --pid 12345
# Check all threads
gdb-cli threads -s b2c3d4
# Get all backtraces
gdb-cli thread-apply -s b2c3d4 bt --all
```
## Best Practices
- Always read source code before drawing conclusions from variable values
- Use `--range` for pagination on large thread counts or deep backtraces
- Use `ptype` to understand complex data structures before examining values
- Check all threads for multi-threaded issues
- Cross-reference types with source code definitions
## Security & Safety Notes
- This skill requires GDB access to processes and core dumps
- Attaching to processes may require appropriate permissions (sudo, ptrace_scope)
- Core dumps may contain sensitive data - handle with care
- Only debug processes you have authorization to analyze
## Related Skills
- `@systematic-debugging` - General debugging methodology
- `@test-driven-development` - Write tests before implementation
## Links
- **Repository**: https://github.com/Cerdore/gdb-cli
- **PyPI**: https://pypi.org/project/gdb-cli/
- **Documentation**: https://github.com/Cerdore/gdb-cli#readme
## Limitations
- Use this skill only when the task clearly matches the scope described above.
- Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
- Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.
Imported SKILL.md format for Claude Code, Codex and Cursor. Review the complete instructions and required tools before use. Package integrity checked; individual agent behavior has not been tested.