The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation section instructs the user to pipe a remote script directly into the shell (curl -sL myopenclawhub.com/install | bash), which allows for arbitrary code execution from a domain that is not a trusted vendor or well-known service.
[EXTERNAL_DOWNLOADS]: The oa management tool is designed to download and install various external components, including Node.js binaries and multiple AI CLI tools (Claude Code CLI, Gemini CLI, etc.), from remote servers without signature verification or version pinning.
[PRIVILEGE_ESCALATION]: The skill provides instructions to disable Android system-level protections, specifically the "Phantom Process Killer" and battery optimizations, which allows background processes to bypass standard OS resource management and safety constraints.
[INDIRECT_PROMPT_INJECTION]: The skill manages external agent definitions and workspace data, creating a vulnerability surface where malicious instructions embedded in that data could influence the agent's behavior.
Ingestion points: Data ingested through oa --restore and files located in ~/.openclaw/.
Boundary markers: None identified to differentiate between data and instructions.
Capability inventory: The skill possesses shell execution, file system access, and network communication capabilities.
Sanitization: No input validation or sanitization mechanisms are described for processed data.

openclaw-android-setup