The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (HIGH): The installation section provides instructions to download a Python script using curl from https://raw.githubusercontent.com/platformio/platformio-core-installer/master/get-platformio.py and execute it immediately. Since the platformio organization is not in the trusted list provided in the security skill, this constitutes unverified remote code execution.
[COMMAND_EXECUTION] (MEDIUM): The skill requires the Bash tool and documentation describes the use of extra_scripts (e.g., pre:scripts/pre_build.py). This feature allows for the execution of arbitrary Python code within the agent's environment during the build phase of a project.
[EXTERNAL_DOWNLOADS] (MEDIUM): The skill facilitates downloading external libraries via lib_deps and pio pkg install. These can point to arbitrary Git URLs or file paths, which could be used to pull malicious code into the development environment.
[PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection because it ingests untrusted project data. Evidence Chain: 1. Ingestion points: Reads platformio.ini and source code files via Read and Grep. 2. Boundary markers: None present to distinguish project data from instructions. 3. Capability inventory: Full Bash access and Python execution through build scripts. 4. Sanitization: None specified for file content before processing.

platformio