sonarcloud-tools
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The skill invokes
pnpm sonaranduv run .claude/skills/sonarcloud-tools/scripts/sonarcloud.py. This involves executing subprocesses and local scripts which could perform arbitrary actions depending on the script content. - [CREDENTIALS_UNSAFE] (MEDIUM): The tool explicitly requires the
SONAR_TOKENenvironment variable. While it mentions 1Password injection, the requirement for a long-lived secret in the environment is a security risk if the environment is compromised. - [DATA_EXFILTRATION] (LOW): The skill performs network operations to SonarCloud APIs. While these are expected for the skill's purpose, they represent a path for data to leave the local environment.
- [INDIRECT_PROMPT_INJECTION] (MEDIUM):
- Ingestion points: The script
sonarcloud.pyretrieves issues, code smells, and hotspots from SonarCloud API (SKILL.md). - Boundary markers: None specified in the provided documentation to delimit API responses from agent instructions.
- Capability inventory: Capability to execute shell commands (
pnpm) and run Python scripts (uv). - Sanitization: No sanitization of the SonarCloud API output is mentioned. Maliciously crafted content within the scanned codebase (e.g., in code comments or issue descriptions) could attempt to influence the agent's subsequent reasoning.
Audit Metadata