Compare two AI tools side by side
Put any two skills or MCP servers next to each other: risk grade, what each can access, how popular it is, and how to install it.
| What | Audrey | ghidra-mcp |
|---|---|---|
| Risk grade | CC · Moderate risk signals Based on public adoption and transparency signals as of Oct 24, 2026. Our opinion, not a code audit or security guarantee. | BB · Some risk signals Based on public adoption and transparency signals as of Oct 16, 2026. Our opinion, not a code audit or security guarantee. |
| Type | App connector | App connector |
| Est. context load | ~500 tokens / turn (~3 tool schemas, 0.25% of 200k context) | ~750 tokens / turn (~4 tool schemas, 0.38% of 200k context) |
| GitHub stars | 26 | 6k |
| Weekly npm downloads | 197 | — |
| Local files | No local folder access noted | No local folder access noted |
| Command execution | Runs shell or system commands | No shell command runner noted |
| Network reach | Connects to outside service API | Connects to outside service API |
| Credentials | No API token required | Requires API key or token |
| Package source | npm (audrey) | GitHub (bethington/ghidra-mcp) |
| Install command | npx -y audrey | Source repo only |
| Bottom line | Audrey is distributed via npm (audrey); review its source code, dependencies, and requested permissions before use. | ghidra-mcp (MCP server) requires careful permission scoping — data exfiltration: Reverse engineering projects often contain proprietary or sensitive binary code; AI agents with unrestricted access could leak analysis to external logs or training data. |
Based on public adoption and transparency signals as of Oct 29, 2026. Our opinion, not a code audit or security guarantee.
Go deeper
Want to inspect exact published files, install hooks and network hosts?
Run a pre-install check on any package, MCP config or SKILL.md, or open a full report with PDF export.