dx-code-analyzer-run
Warn
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing shell commands via the Salesforce CLI (sf code-analyzer). The bundled script
scripts/describe-rule.jsconstructs shell command strings using input arguments without sanitization, creating a potential command injection vector. - [COMMAND_EXECUTION]: In contrast, the
scripts/list-rules.jsutility implements a robust validation mechanism for input tokens against a whitelist, indicating an inconsistent security implementation within the skill's own tools. - [REMOTE_CODE_EXECUTION]: The skill uses Node.js to execute local scripts like
apply-fixes.jswhich modify project source code by applying transformations extracted from scan results. - [PROMPT_INJECTION]: The instructions in
SKILL.mdcontain strict directives that mandate the use of bundled scripts and explicitly forbid the use of common alternatives like jq, Python, or standard agent tools. - [DATA_EXFILTRATION]: The skill reads source code and generates local report files. While no exfiltration to untrusted domains is detected, the skill processes sensitive project data and instructions encourage users to share reports for review.
Audit Metadata