The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches a configuration and setup script from the official KernelSU-Next GitHub repository (github.com/KernelSU-Next). These downloads are expected as they are part of the core functionality for integrating the specified root solution.
[REMOTE_CODE_EXECUTION]: Remote scripts downloaded from the KernelSU-Next project are piped directly to bash for execution in REFERENCE.md and scripts/build_gki_ksun.sh. This is the documented installation method for this tool.
[COMMAND_EXECUTION]: The skill includes shell scripts (scripts/build_gki_kernel.sh and scripts/build_gki_ksun.sh) that execute system commands such as repo, git, and build utilities (build.sh, bazel) using arguments provided during the agent's workflow.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection via user-supplied data.
- Ingestion points: Arguments for manifest branches, build configuration paths, and Bazel targets are accepted by the helper scripts in the scripts/ directory.
- Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat these external inputs as potentially untrusted data.
- Capability inventory: The skill possesses capabilities to modify the local filesystem, perform network operations via repo and curl, and execute shell commands.
- Sanitization: The provided scripts perform minimal validation on user-provided inputs before incorporating them into shell command executions.

gki-kernel