diff --git a/data/tools/rea.yaml b/data/tools/rea.yaml new file mode 100644 index 0000000..7e9698c --- /dev/null +++ b/data/tools/rea.yaml @@ -0,0 +1,43 @@ +name: rea +binary: rea +aliases: +- rea-agents +category: +- security +- debug +- mcp +lang: +- all +summary: Local CLI and MCP tools for agents to inspect binaries and JavaScript/Electron applications with evidence provenance. +homepage: https://rea.tools +docs: https://github.com/morluto/rea/blob/main/docs/cli.md +detect: + version_args: + - --version +use_when: +- Inspect a caller-selected application artifact and consume JSON observations, source locations, and explicit unknowns +- Map readable JavaScript/Electron modules, imports, and IPC relationships before choosing deeper analysis +avoid_when: +- Deep native analysis is required but no supported Hopper, Ghidra, or IDA provider is configured +- Opaque JavaScript bytecode requires a decoder that is not available +- The workflow assumes every command is read-only or automatically sandboxed +risk: + level: high + effects: + - read_files + - write_files + - read_processes + - execute_code + - network_access + - install_packages + requires_auth: false + destructive: false + confirmation_required_for: + - launching or injecting into a target process + - changing analysis annotations + - installing Hopper or changing agent configuration during setup +guardrails: +- Select the exact artifact, process, and output paths; use static inspection when execution is not authorized. +- Configure bring-your-own Hopper, Ghidra, or IDA for deep native analysis; do not infer engine availability from MCP startup. +- Use annotation and runtime operations only when the requested task authorizes their effects. +- Review the setup plan before approving configuration writes or the optional Hopper installation.