lightcraft-cli-library
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches platform-specific binaries for the Lightcraft CLI from GitHub. The skill implements a robust verification mechanism using a "runtime.lock.json" file that defines trusted URLs and mandatory SHA256 hashes for both archives and binaries.
- [COMMAND_EXECUTION]: Runs the Lightcraft CLI using the "subprocess" module. The implementation explicitly uses argument lists and avoids shell execution, which effectively mitigates the risk of command injection.
- [REMOTE_CODE_EXECUTION]: Downloads and executes a native binary as its primary function. This behavior is considered safe because it is the intended purpose of the skill, and the process is protected by multi-stage integrity checks (SHA256 verification of the archive and the extracted binary) and domain whitelisting.
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted image metadata and file paths. It defends against injection attacks by using strict JSON schema validation for all command inputs and enforcing integrity boundaries via resource-pinning on plan execution.
- [DYNAMIC_EXECUTION]: Dynamically loads internal utility modules (e.g., "command_gateway.py", "photo_workflows.py") using Python's "importlib". This is used for code modularity and is restricted to the skill's own local directory, which avoids risks associated with loading arbitrary external code.
Audit Metadata