gcc
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
subprocess.runandsubprocess.Popento execute external binaries such ascmakeand various tools from the ARM GNU Toolchain (e.g.,arm-none-eabi-size). - Evidence found in
scripts/gcc_build.py(actions:configure,build,clean) andscripts/gcc_size.py(analyze,compare). - While the skill uses argument lists to mitigate shell injection, the paths to these executables and their arguments are derived from user configuration and project files.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from the workspace, creating an attack surface for indirect prompt injection.
- Ingestion points: The skill reads and parses
CMakeLists.txt,CMakePresets.json, linker scripts, and the output of build tools across all scripts (specificallyscripts/gcc_project.py,scripts/gcc_build.py, andscripts/gcc_size.py). - Boundary markers: Absent. No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content embedded in these files.
- Capability inventory: The skill possesses the capability to execute shell commands and write log files to the workspace.
- Sanitization: Absent. Processing relies on regular expressions and standard JSON parsing, which may not account for adversarial content designed to influence agent behavior.
Audit Metadata