### Standard First-Time Setup and Deploy

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Use this sequence for initial setup on a new machine and to deploy the plugin. It installs dependencies, builds the plugin, and copies it to the Ghidra profile.

```bash
python -m tools.setup ensure-prereqs --ghidra-path "C:\\ghidra_12.1_PUBLIC"
```

```bash
python -m tools.setup build
```

```bash
python -m tools.setup deploy --ghidra-path "C:\\ghidra_12.1_PUBLIC"
```

--------------------------------

### Deploy Project with Setup CLI

Source: https://github.com/bethington/ghidra-mcp/blob/main/tools/README.md

Execute the deploy command using the setup CLI, providing the path to your Ghidra installation. This is part of the deployment process for the project.

```bash
python -m tools.setup deploy         --ghidra-path F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Recommended Script Order: Fresh Start

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/QUICK_REFERENCE_SCRIPTS.md

For a new project or initial setup, run these scripts in the specified order. 'BuildHashIndex_ProjectFolder.java' creates the initial documentation index, 'FixSymbolConflicts_ProjectFolder.java' resolves naming issues, and 'BatchPropagateToAllVersions_ProjectFolder.java' applies documentation across all project versions.

```java
1. BuildHashIndex_ProjectFolder.java
   ├─ Choose: "Start Fresh" (for first run)
   └─ Output: ~/ghidra_function_hash_index.json

2. FixSymbolConflicts_ProjectFolder.java (optional)
   ├─ Choose: "Fix Conflicts Only" (if conflicts found)
   └─ Or "Cancel" (if no conflicts)

3. BatchPropagateToAllVersions_ProjectFolder.java
   ├─ Choose: "All Binaries in Project"
   └─ Propagates to all versions
```

--------------------------------

### Verify Gradle version task example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/GRADLE_PHASE1_BACKLOG.md

Provides example commands to run the `verifyVersion` task. This task compares the Ghidra version from `pom.xml` with the version reported by the Ghidra installation. It exits cleanly even if `GHIDRA_INSTALL_DIR` is not set.

```bash
./gradlew verifyVersion
./gradlew verifyVersion -PGHIDRA_INSTALL_DIR=F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Show All Setup CLI Subcommands

Source: https://github.com/bethington/ghidra-mcp/blob/main/tools/README.md

Use this command to display all available subcommands for the setup CLI. This is useful for understanding the full range of project setup, build, deploy, and version management operations.

```bash
python -m tools.setup --help
```

--------------------------------

### Prepare Ghidra classpath task example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/GRADLE_PHASE1_BACKLOG.md

Demonstrates running the `prepareGhidraClasspath` task, which verifies the existence of 18 required Ghidra jars. Gradle uses `fileTree` directly for compilation, eliminating the need for `mvn install:install-file`.

```bash
./gradlew prepareGhidraClasspath -PGHIDRA_INSTALL_DIR=F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Build and Start Single GhidraMCP Instance

Source: https://github.com/bethington/ghidra-mcp/blob/main/docker/README.md

Build the Docker image and start a single instance of the GhidraMCP headless server using docker-compose. Use this for initial setup and testing.

```bash
# Build and start
cd docker
docker-compose up -d

# Check status
curl http://localhost:8089/check_connection
```

--------------------------------

### Gradle Start Ghidra Command

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/GRADLE_PHASE1_BACKLOG.md

This command starts Ghidra using Gradle, integrated into the setup CLI.

```python
cmd_start_ghidra → run_gradle(..., ['startGhidra'])
```

--------------------------------

### Setup Commands with Gradle Backend

Source: https://github.com/bethington/ghidra-mcp/blob/main/CLAUDE.md

Use these commands when `TOOLS_SETUP_BACKEND` is set to `gradle`. This routes the same setup commands to a Gradle backend.

```bash
$env:TOOLS_SETUP_BACKEND = "gradle"
python -m tools.setup build
python -m tools.setup preflight --ghidra-path F:\ghidra_12.1_PUBLIC
python -m tools.setup deploy    --ghidra-path F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Start Ghidra Application (macOS)

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Launches the Ghidra application. After starting, enable the GhidraMCP plugin via Tools > GhidraMCP > Start MCP Server.

```bash
/opt/homebrew/opt/ghidra/libexec/ghidraRun
```

--------------------------------

### Install Prerequisites (macOS)

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Installs necessary tools like OpenJDK, Maven, Python, and Ghidra using Homebrew.

```bash
brew install openjdk@21 maven python ghidra
```

--------------------------------

### Add Example Usage for New Tool

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

Create an example Python script in the `examples/` directory to demonstrate how to use the newly added MCP tool.

```python
# examples/use-new-tool.py
from bridge_mcp_ghidra import my_new_tool

result = my_new_tool("example", param2=5)
print(result)
```

--------------------------------

### Running Ghidra Preflight Check

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

Execute the preflight check for Ghidra setup, specifying the path to your Ghidra installation.

```bash
# 2. Run preflight against your Ghidra install
python -m tools.setup preflight --ghidra-path "C:\path\to\ghidra"
```

--------------------------------

### Start fun-doc Dashboard Only

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/README.md

Use this command to start only the web dashboard without initiating any documentation workers.

```bash
python fun_doc.py --no-worker
```

--------------------------------

### Start Web Dashboard Separately

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/README.md

This command starts only the web dashboard. It requires that fun_doc.py is already running to connect to the event bus.

```bash
python web.py
```

--------------------------------

### String Labeling Example with Suffixes

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/STRING_LABELING_CONVENTION.md

Illustrates how to apply labels to strings, including examples of duplicate handling and descriptive suffixes like Fmt, Msg, Title, Dir, Alt, Upper, Lower, and Local.

```text
szOpt_SmallOff      # First instance
szOpt_SmallOff2     # Second instance
szOpt_SmallOff3     # Third instance
```

```text
Suffix | When to Use | Example
-------|-------------|---------
`Fmt` | Format strings | `szDbg_TileCacheFmt`
`Msg` | Message body | `szSecurity_BufferOverrunMsg`
`Title` | Dialog title | `szSecurity_BufferOverrunTitle`
`Dir` | Directory path | `szPath_MonstersDir`
`Alt` | Alternative version | `szApi_MessageBoxA_Alt`
`Upper` | Uppercase variant | `szDllName_Kernel32Upper`
`Lower` | Lowercase variant | `szDll_D2GameLower`
`Local` | Local/embedded copy | `szDllName_D2ClientLocal`
```

--------------------------------

### Start fun-doc Dashboard and Worker Loop

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/README.md

This command starts the main entry point for fun-doc, launching both the web dashboard and the idle worker loop for function documentation.

```bash
python fun_doc.py
```

--------------------------------

### Install Debugger Dependencies and Run Debugger

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Install optional Windows-only Python dependencies for the standalone debugger server and then run the debugger.

```text
python -m pip install -r requirements-debugger.txt
python -m debugger
```

--------------------------------

### Install Ghidra Dependencies on Windows

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Installs Ghidra dependencies, ensuring all necessary JARs are available. Specify the Ghidra installation path using the --ghidra-path argument.

```text
# Windows
python -m tools.setup install-ghidra-deps --ghidra-path "C:\path\to\ghidra_12.1_PUBLIC"
```

--------------------------------

### Feature Request Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

An example of how to structure a feature request, including title, use case, proposed solution, and priority.

```text
Title: Add tool to analyze register usage patterns

Use Case:
Need to find functions that don't follow calling convention
for obfuscation analysis

Proposed Solution:
New tool: analyze_register_usage(function_name)
Returns: {"register_changes": {...}, "violations": [...]} 

Priority: Medium - helpful for malware analysis
```

--------------------------------

### Ensuring Prerequisites Installation

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

Command to ensure all necessary prerequisites for Ghidra-MCP are installed, including specifying the Ghidra path.

```bash
# 3. Install prerequisites
python -m tools.setup ensure-prereqs --ghidra-path "C:\path\to\ghidra"
```

--------------------------------

### Start fun-doc with a Specific Provider

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/README.md

This command starts fun-doc and specifies a particular LLM provider, such as 'claude', to be used for documentation tasks.

```bash
python fun_doc.py --provider claude
```

--------------------------------

### Install System Prerequisites on Linux (Ubuntu/Debian)

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Installs essential system packages including OpenJDK, Maven, Python, pip, curl, jq, and unzip required for building Ghidra MCP on Ubuntu/Debian-based systems.

```bash
sudo apt update && sudo apt install -y openjdk-21-jdk maven python3 python3-pip curl jq unzip
```

--------------------------------

### Bug Report Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

An example of how to structure a bug report, including title, environment, steps to reproduce, and expected vs. actual behavior.

```text
Title: Decompile timeout on binary > 10MB

Environment:
- OS: Windows 11
- Ghidra: 12.1
- Binary: 12MB x86-64

Steps:
1. Load 12MB binary in Ghidra
2. Wait for analysis
3. Call decompile_function("main")

Error: RequestTimeout after 30 seconds

Expected: Decompilation completes or returns gracefully
```

--------------------------------

### Renaming Existing Files Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/NAMING_CONVENTIONS.md

Illustrates renaming of existing files to comply with recommended conventions during migration.

```text
CLAUDE.md
ordinal_fix_log_*.txt
Old reports
Examples .py files
Docs in docs/
```

--------------------------------

### Bump Project Version with Setup CLI

Source: https://github.com/bethington/ghidra-mcp/blob/main/tools/README.md

Use the bump-version command with the setup CLI to update the project's version number. Specify the new version using the --new flag.

```bash
python -m tools.setup bump-version   --new 5.10.0
```

--------------------------------

### File Layout Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/benchmark/regression/__schema__.md

Illustrates the directory structure for regression assertion files.

```text
fun-doc/benchmark/regression/
  __schema__.md            (this file)
  Benchmark.dll.yaml       (assertions for /testing/benchmark/Benchmark.dll)
  BenchmarkDebug.exe.yaml  (assertions for /testing/benchmark/BenchmarkDebug.exe)
```

--------------------------------

### Field Offset Verification Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Example demonstrating how to verify field offsets by comparing structure definitions with actual assembly access. This helps ensure type accuracy.

```text
Structure Definition          Assembly Access
Offset 0x0: dwType (uint)     MOV EAX, [pUnit + 0x0] ✓
Offset 0x4: dwFlags (uint)    CMP [pUnit + 0x4], 0   ✓
Offset 0x8: pNext (UnitAny*)  LEA ECX, [pUnit + 0x8] ✓
```

--------------------------------

### Structure Alignment Verification Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Example illustrating how to verify structure alignment assumptions based on its fields. This is important for correct memory layout and access.

```text
If structure has:
  DWORD (4 bytes)
  DWORD (4 bytes)
  WORD (2 bytes)
  WORD (2 bytes)

Expected alignment: 4-byte (DWORD-aligned)
Expected size: 12 bytes (3 DWORDs)
```

--------------------------------

### Structure Alignment and Padding Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

An example detailing the identification of padding and alignment within a structure based on offset analysis. This helps in understanding the memory layout and potential unused bytes.

```text
Offset 0x00-0x03: dwType (uint) - 4 bytes
Offset 0x04-0x07: dwFlags (uint) - 4 bytes
Offset 0x08-0x0B: [UNKNOWN] - 4 bytes
Offset 0x0C-0x0D: wHealth (ushort) - 2 bytes
Offset 0x0E-0x0F: [PADDING/ALIGNMENT] - 2 bytes
```

--------------------------------

### Starting Standalone Debugger Server

Source: https://github.com/bethington/ghidra-mcp/blob/main/CLAUDE.md

Command to launch the standalone debugger server, which will listen on port 8099.

```bash
python -m debugger
```

--------------------------------

### Build and Run Ghidra-MCP Docker Container

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Builds and starts the Ghidra-MCP headless server in detached mode using Docker Compose. This is the quickest way to get the server running.

```bash
# Build and run
docker-compose up -d ghidra-mcp
```

--------------------------------

### Setup Commands with Maven Backend (Default)

Source: https://github.com/bethington/ghidra-mcp/blob/main/CLAUDE.md

Default commands routed through the Maven backend. Includes build, preflight checks, dependency verification, and deployment.

```text
python -m tools.setup build
python -m tools.setup preflight      --ghidra-path F:\ghidra_12.1_PUBLIC
python -m tools.setup ensure-prereqs --ghidra-path F:\ghidra_12.1_PUBLIC
python -m tools.setup deploy         --ghidra-path F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Recommended Script Order: Updates

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/QUICK_REFERENCE_SCRIPTS.md

When updating documentation after initial setup, follow this order. 'BuildHashIndex_ProjectFolder.java' merges new documentation into the existing index, and 'BatchPropagateToAllVersions_ProjectFolder.java' updates other versions from the current program.

```java
1. BuildHashIndex_ProjectFolder.java
   ├─ Choose: "Merge with Existing"
   └─ Updates index with new documentation

2. BatchPropagateToAllVersions_ProjectFolder.java
   ├─ Choose: "Current Program Only"
   └─ Updates all other versions from current
```

--------------------------------

### Build Project

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/releases/RELEASE_CHECKLIST.md

Compile the project. This is a standard build step.

```bash
python -m tools.setup build
```

--------------------------------

### Installing Debugger Dependencies

Source: https://github.com/bethington/ghidra-mcp/blob/main/CLAUDE.md

Optional command to install dependencies required for the debugger.

```bash
python -m pip install -r requirements-debugger.txt
```

--------------------------------

### Install Ghidra Dependencies into Maven (macOS)

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Installs Ghidra JARs into the local Maven repository using a Python script. Ensure the --ghidra-path is correctly set to your Ghidra installation directory.

```bash
python -m tools.setup install-ghidra-deps \
   --ghidra-path /opt/homebrew/opt/ghidra/libexec
```

--------------------------------

### Log File Naming Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/NAMING_CONVENTIONS.md

Example of a log file name adhering to the snake_case_date format.

```text
build_log_20251105.txt
```

--------------------------------

### Build the Benchmark Binary

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/benchmark/README.md

Compile the fun-doc benchmark binary, which produces build/Benchmark.dll. This is a prerequisite for real mode.

```bash
python fun-doc/benchmark/build.py
```

--------------------------------

### Setup Ghidra Benchmark Integration

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/benchmark/README.md

Import the compiled Benchmark.dll into a running Ghidra instance. This step is idempotent and can be run multiple times. The default import path is /testing/benchmark/Benchmark.dll, but can be overridden with FUNDOC_BENCHMARK_PROGRAM.

```bash
python fun-doc/benchmark/setup_ghidra_benchmark.py
```

--------------------------------

### Gradle Install Ghidra Dependencies Command

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/GRADLE_PHASE1_BACKLOG.md

This command installs Ghidra classpath dependencies using Gradle.

```python
cmd_install_ghidra_deps → run_gradle(..., ['prepareGhidraClasspath'])
```

--------------------------------

### Maven Build Process Overview

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/MAVEN_VERSION_MANAGEMENT.md

This outlines the steps involved in the Maven build process for version substitution. It highlights how Maven reads the `pom.xml` version and substitutes it into resource files.

```text
1. Developer updates pom.xml: <version>X.Y.Z</version>
2. Maven build runs: mvn clean package assembly:single
3. Maven processes resources:
   - Reads pom.xml version
   - Substitutes ${project.version} in version.properties
   - Substitutes ${project.version} in extension.properties
4. Java plugin loads version.properties at runtime
5. All APIs report correct version automatically
```

--------------------------------

### Example Global Names

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/prompts/fix-globals.md

Examples of valid global variable names following the required naming convention.

```text
g_dwPlayerCount
g_pMainUnit
g_szConfigPath
g_pfnCallback
```

--------------------------------

### Run preflight check task example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/GRADLE_PHASE1_BACKLOG.md

Shows how to execute the `preflight` task, which validates the Java environment, Ghidra jar presence, and write access to extension directories. This task depends on `verifyVersion`.

```bash
./gradlew preflight -PGHIDRA_INSTALL_DIR=F:\ghidra_12.1_PUBLIC
```

--------------------------------

### Install Test Requirements

Source: https://github.com/bethington/ghidra-mcp/blob/main/docker/README.md

Install the necessary Python packages for running integration tests against the GhidraMCP headless server.

```bash
# Install test requirements
pip install -r tests/requirements.txt
```

--------------------------------

### Initial Provider Policy Implementation

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/FUNCTION_DOCUMENTATION_SCALE_RFC.md

Outlines the first steps for implementing provider policies, focusing on leveraging existing configurations and adding mode-specific defaults.

```text
use existing provider configuration
add per-mode defaults where needed
```

--------------------------------

### Worker Started Acknowledgment

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/templates/dashboard.html

Handles the 'worker_started_ack' event from the socket. This event typically signifies that a worker has successfully started or resumed.

```javascript
socket.on('worker_started_ack', d => {
  /* pane created by worker_started event */
});
```

--------------------------------

### Field Offset Verification Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Example of verifying field offsets for specific functions, indicating successful access with a checkmark.

```markdown
ProcessUnit: [pUnit + 0x0] = dwType ✓
ValidateUnit: [pUnit + 0x4] = dwFlags ✓
ApplyDamage: [pUnit + 0xC] = wHealth ✓
```

--------------------------------

### Complete Example: String Type Application

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/TOOL_USAGE_GUIDE.md

Demonstrates the full process of applying a string data type, renaming it with Hungarian notation, and the expected return messages for each step.

```python
# Address: 0x0040BC08, Data: "VIDEO" string (6 bytes)

# Step 3a: Apply type
apply_data_type("0x0040bc08", "char[6]")
# Returns: "Successfully applied data type 'char[6]' at 0x0040bc08 (size: 6 bytes)"

# Step 3b: Rename with Hungarian notation
rename_or_label("0x0040bc08", "szVideoSection")
# Returns: "Success: Renamed defined data at 0x0040bc08 to 'szVideoSection'"
```

--------------------------------

### Endpoint Metadata Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/HEADLESS_PARITY_RFC.md

Example JSON structure for an endpoint entry in the compatibility matrix, including headless status and requirements.

```json
{
  "path": "/decompile_function",
  "method": "GET",
  "category": "decompile",
  "params": ["address", "program"],
  "description": "Decompile function",
  "compatibility": {
    "gui": true,
    "headless": true,
    "headless_status": "full",
    "reason": "",
    "workflows": ["function_docs", "batch_docs"],
    "requirements": {
      "requires_project": true,
      "requires_program": true,
      "requires_selection": false,
      "requires_codebrowser": false,
      "requires_gui_tool": false,
      "writes_project": false
    }
  }
}
```

--------------------------------

### Platform Plan: First Backend

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/project-management/DYNAMIC_KNOWLEDGE_VALIDATION_RFC.md

Details the platform plan, starting with a cross-platform design contract and implementing the first backend for Windows/dbgeng targeting PE files. This establishes the initial environment for dynamic validation.

```text
Design contract
  debugger-agnostic evidence model and workflow semantics.

First backend
  Windows/dbgeng for PE targets.

Later backends
  Linux/gdb and macOS/lldb.
```

--------------------------------

### Example 3: Helper Structure Creation for Skill Execution

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Shows the creation of a helper structure, SkillObject, for tracking runtime skill execution state. Useful for managing temporary data related to function calls.

```text
SkillObject (48 bytes):
  - dwType: uint (skill type identifier)
  - pOwnerUnit: UnitAny * (unit executing skill)
  - pTargetUnit: UnitAny * (target of skill)
  - dwSkillId: uint (skill index)
  - dwLevel: uint (skill level)
  - dwCharges: uint (remaining uses)
  - pSkillData: SkillData * (execution parameters)
```

--------------------------------

### Verify Ghidra Installation Directory in Docker

Source: https://github.com/bethington/ghidra-mcp/blob/main/docker/README.md

Execute a command inside the Ghidra-MCP Docker container to verify the Ghidra installation path.

```bash
docker exec ghidra-mcp ls /opt/ghidra
```

--------------------------------

### Example 1: Identifying UnitAny Parameters

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Demonstrates identifying a UnitAny structure by analyzing function parameters, assembly access patterns, and cross-referencing similar functions. Useful for understanding game entity structures.

```text
Function ProcessUnit (0x401000):
  Parameter 1: int * pUnit  # Generic type

Assembly:
  MOV EAX, [pUnit + 0x0]    # Read at offset 0x0
  CMP [pUnit + 0x4], 0      # Compare at offset 0x4
```

```text
UnitAny * pUnit
  - dwType at offset 0x0
  - dwFlags at offset 0x4
  - wHealth at offset 0xC
  - ... (248 more bytes of game entity data)
```

--------------------------------

### Commit Message Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

An example of a commit message following the conventional format, including type, scope, subject, body, and footer.

```git
feat(tools): Add batch_analyze_structures tool

- Implements bulk structure discovery
- 90% API call reduction vs individual calls
- Tested on 3 large binaries

Closes #123
```

--------------------------------

### Invalid Plate Comment Examples

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/PLATE_COMMENT_BEST_PRACTICES.md

Examples of plate comments that fail validation due to length, missing steps, or missing sections.

```text
Process data.

Parameters:
- pContext: Context pointer

Returns: Success code
```

```text
Algorithm:
Initialize variables, validate inputs, process data, return result.

Parameters:
- pContext: Context pointer

Returns:
- 0 on success
```

```text
Algorithm:
1. Initialize
2. Process
3. Return

This function processes data.
```

--------------------------------

### High-Xref Bitfield Global Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/prompts/step-globals.md

An example of a detailed plate-comment for a high-xref bitfield global, including 'Set by', 'Read by', 'Source', and 'Bitfield' sections.

```text
Bitmap of currently-active quests for the player; bit N = quest N active.

Set by: ProcessQuestUpdate, InitQuestState
Read by: RenderQuestLog, IsQuestActive (and 14 other readers)
Source: QuestC.cpp:0x14
Bitfield:
  0x0001 = QUEST_DENOFEVIL
  0x0002 = QUEST_SISTERS_BURIAL

```

--------------------------------

### Build and Deploy Ghidra MCP (macOS)

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Ensures prerequisites are met, builds the project, and deploys the extension. The extension will be installed to ~/Library/ghidra/ghidra_12.1_PUBLIC/Extensions/GhidraMCP/.

```bash
python -m tools.setup ensure-prereqs \
   --ghidra-path /opt/homebrew/opt/ghidra/libexec
python -m tools.setup build
python -m tools.setup deploy \
   --ghidra-path /opt/homebrew/opt/ghidra/libexec
```

--------------------------------

### Install Ghidra Maven Dependencies on Linux

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Installs only the Ghidra JAR dependencies into the local Maven repository (`~/.m2/repository`) without building or deploying the extension.

```bash
python -m tools.setup install-ghidra-deps --ghidra-path ~/ghidra_12.1_PUBLIC
```

--------------------------------

### Type Application Summary Example

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

A summary format for documenting type application changes, including the number of functions updated, parameter, old type, new type, and verification details.

```markdown
Total functions updated: 12
Parameter: pUnit
Old type: int *
New type: UnitAny *
Verification: All functions access fields at 0x0-0xF8
```

--------------------------------

### Verify VC6 cl.exe Installation

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/benchmark/tools/vc6/README.md

Run this command to check if the Visual C++ 6.0 compiler (cl.exe) is correctly installed and accessible in the expected path.

```bash
./tools/vc6/VC98/Bin/cl.exe
```

--------------------------------

### Example 2: Creating SkillTableEntry Structure

Source: https://github.com/bethington/ghidra-mcp/blob/main/docs/prompts/DATA_TYPE_INVESTIGATION_WORKFLOW.md

Illustrates the process of creating a new structure, SkillTableEntry, based on offset analysis from multiple functions. Useful when existing structures do not match observed data access patterns.

```asm
LEA ECX, [pSkillTable + EAX*0x23C + 0x1A4]  # Stride 0x23C, offset 0x1A4
MOV EDX, [ECX]                               # Read at +0x1A4
CMP [ECX + 0x4], 0                           # Compare at +0x1A8
```

```python
create_struct("SkillTableEntry", [
    {"name": "dwType", "type": "uint"},
    {"name": "dwId", "type": "uint"},
    # ... 100 more bytes of fields
    {"name": "dwShiftAmount", "type": "byte", "offset": 0x1A4},
    {"name": "dwWeaponCoeff", "type": "byte", "offset": 0x1A5},
    {"name": "dwBaseDamage", "type": "uint", "offset": 0x1A8},
    # ... more fields
])
```

--------------------------------

### Preview Markdown Documentation

Source: https://github.com/bethington/ghidra-mcp/blob/main/CONTRIBUTING.md

Display the content of the endpoints.json file using the 'more' command for preview.

```bash
cat tests/endpoints.json | more
```

--------------------------------

### Worker Started Notification

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/templates/dashboard.html

Adds an activity log entry when a worker starts, including details like provider, restoration status, and type (auto or count-based).

```javascript
socket.on('worker_started', d => {
  addActivity(`Worker ${d.worker_id}: ${d.provider} ${d.restored ? 'restored ' : ''}${d.continuous ? 'Auto' : 'x' + d.count}`, 'scan');
});
```

--------------------------------

### /create_project

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Create a Ghidra project.

```APIDOC
## POST /create_project

### Description
Create a Ghidra project.

### Method
POST

### Endpoint
/create_project
```

--------------------------------

### Handle Worker Started Event

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/templates/dashboard.html

Creates or retrieves a worker pane and appends a 'Worker started' message when a worker begins. It also ensures the 'Stop All' button is visible.

```javascript
socket.on("worker_started", d => {
  getOrCreatePane(d.worker_id, d.provider, d.binary || '');
  appendToPane(
    d.worker_id,
    d.restored ? `Worker restored after restart: ${d.provider} ${d.continuous ? 'Auto' : 'x' + d.count}` : `Worker started: ${d.provider} ${d.continuous ? 'Auto' : 'x' + d.count}`,
    'scan'
  );
  // Show Stop All button when any worker is running
  const stopAllBtn = document.getElementById('btnStopAll');
  if (stopAllBtn) stopAllBtn.style.display = '';
});
```

--------------------------------

### Run Benchmark in Mock Mode (Default)

Source: https://github.com/bethington/ghidra-mcp/blob/main/fun-doc/benchmark/README.md

Execute the benchmark in mock mode with default settings for fast tier and baseline variant, comparing against the latest run.

```bash
python fun-doc/benchmark/run_benchmark.py --mock --tier fast --compare
```

--------------------------------

### Build Ghidra MCP Project

Source: https://github.com/bethington/ghidra-mcp/blob/main/README.md

Commands to set up prerequisites, build, and deploy the Ghidra MCP project using the Python-first workflow. Ensure Ghidra is installed and its path is correctly specified.

```bash
# Recommended: direct Python-first workflow
python -m tools.setup ensure-prereqs --ghidra-path "C:\\ghidra_12.1_PUBLIC"
python -m tools.setup build
python -m tools.setup deploy --ghidra-path "C:\\ghidra_12.1_PUBLIC"
```