The Agent Skills Directory

[SAFE]: The skill is a static analysis tool designed to review embedded systems code for correctness and compliance with safety standards (ISO 26262, MISRA, AUTOSAR).
[COMMAND_EXECUTION]: The skill instructions include logic to search for a local .git directory to identify the project root. This is used as a heuristic to determine the correct location for saving generated reports in an analysis/ subdirectory.
[DATA_EXFILTRATION]: The skill is configured to save findings to a local HTML file. It does not contain instructions to transmit this data to any remote server or non-whitelisted domain.
[PROMPT_INJECTION]: As the skill is designed to ingest and process arbitrary source code snippets, it has a theoretical surface for indirect prompt injection (Category 8). Malicious instructions could be embedded within code comments to attempt to influence the agent's review. However, the skill's restricted output format and focused purpose make it a low-risk environment for such attacks.

code-review