esp32-firmware-engineer

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The validate.sh script executes a Python script using a hardcoded absolute path: /Users/adamlipecz/.codex/skills/.system/skill-creator/scripts/quick_validate.py. This relies on a specific local file system structure and executes an external file that is not part of the skill's distributed content, posing a risk if the environment is shared.
  • [REMOTE_CODE_EXECUTION]: The validate.sh script uses pip install to download and install the pyyaml package from the internet.
  • [EXTERNAL_DOWNLOADS]: The scripts/check_plugin_compatibility.py script fetches configuration data from Espressif's official GitHub repository (raw.githubusercontent.com/espressif/esp-adf/master/README.md) to verify version compatibility.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as serial logs, firmware source code, and configuration files (e.g., sdkconfig, partitions.csv). It possesses powerful capabilities, including building and flashing firmware using idf.py via shell scripts.
  • Ingestion points: User-provided serial logs and project source code snippets processed as part of debugging and implementation tasks.
  • Boundary markers: Absent; there are no explicit delimiters or warnings to ignore embedded instructions in the processed data.
  • Capability inventory: Subprocess execution of idf.py, git, and shell wrappers (scripts/build.sh, scripts/flash.sh) which interact with the host system and connected hardware.
  • Sanitization: Absent; the skill lacks filtering or validation for external content interpolated into build and flash workflows.
  • [COMMAND_EXECUTION]: Shell scripts in the scripts/ directory (build.sh, flash.sh, monitor.sh, menuconfig.sh) execute system-level commands including idf.py and git via subprocess calls to manage the firmware development toolchain.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 15, 2026, 03:28 AM
Security Audit — agent-trust-hub — esp32-firmware-engineer