Use Hopper

---
name: use-hopper
description: Use Hopper on macOS for disassembly, graphs, pseudocode, Objective-C and Swift presentation, procedure and type edits, extensions, and debugging. Use for document setup, navigation, annotation, evidence, comparison, and discovery.
metadata:
  hermes:
    category: reverse-engineering
    tags: [hopper, disassembly, macos]
---

# Use Hopper

## Overview

Use Hopper as a Mac-native interactive analysis document. Discover the installed edition and capabilities before relying on decompiler, debugger, scripting, SDK, AI, or MCP features.

Read [references/hopper-workflow.md](references/hopper-workflow.md) for capability and project records, Apple handoffs, and official sources.

## Workflow

1. Preserve and preflight.
   - Record artifact hash, format, architecture or slice, UUID or build ID, signing or encryption clues, and question.
   - Use the owning domain skill for Apple bundle, Mach-O, runtime, or symbol context.

2. Discover capabilities.
   - Record Hopper version, edition and license context, installed loaders and CPU backends, decompiler, debugger, Python, SDK extensions, and any AI or MCP surface.
   - Record whether a feature is present, configured, or merely documented for another release.

3. Create the analysis document.
   - Record loader, selected architecture and image, base address, analysis options, and document path.
   - Analyze universal slices separately.

4. Navigate and annotate.
   - Compare assembly, control-flow graph, pseudocode, strings, references, procedures, types, Objective-C and Swift views, and file metadata.
   - Record comments, procedure changes, types, and renames with evidence and confidence.

5. Automate narrowly.
   - Use `$script-hopper-analysis` for Python API or extension work after verifying installed API behavior.
   - Use `$connect-hopper-mcp` for the local MCP data-flow and capability profile.

6. Keep external and dynamic features explicit.
   - Treat LLDB or GDB debugging as a separate runtime stage.
   - Treat AI or MCP use as a separate data-flow decision; record what leaves the document and where it goes.

7. Preserve and compare.
   - Save the Hopper document separately from the original.
   - Preserve screenshots or minimal exports and compare important pseudocode claims with assembly or another tool.

## Guardrails

- Do not enable patching or overwrite the original without an explicit transformation request.
- Do not claim Hopper's Swift or Objective-C presentation is complete source recovery.
- Do not assume an MCP, AI, debugger, loader, or Python surface exists in the installed build.
- Do not transmit artifact data through an external integration without explicit approval.
- Keep interactive document edits separate from scripted and MCP-driven actions so the evidence record identifies their source.

## Output

Return artifact identity, Hopper capability inventory, document and analysis context, observations, annotations, automation or integration data flow, preserved state, and next verification. When the task stops at capability discovery, explicitly report `preflight only; no document created` and leave loader, analysis, decompiler, debugger, and integration behavior unverified.