The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands primarily through the sf (Salesforce CLI) and jq tools. The driver script scripts/integration-driver.sh implements extensive input validation using regular expressions for org aliases, counts, and DeveloperNames to prevent command injection vulnerabilities.
[DATA_EXFILTRATION]: The skill includes a dedicated credential management system. The redact_credentials function in scripts/integration-driver.sh ensures that any generated passwords (e.g., from System.setPassword in Salesforce) are moved to a restricted file with 0600 permissions and replaced with placeholders in all other artifacts and logs to prevent accidental exposure.
[REMOTE_CODE_EXECUTION]: The driver script orchestrates several local 'leaf' scripts. All script paths are resolved relative to the skill's internal directory structure. There is no evidence of the skill downloading and executing arbitrary remote scripts from untrusted sources.
[TIME_DELAYED_CONDITIONAL]: The skill implements a safe_to_write guardrail that queries the Salesforce Organization object. It refuses to proceed with modifications if the target org is detected as a production customer environment (verified via IsSandbox, TrialExpirationDate, and OrganizationType), which is a security best practice for administrative tools.
[DYNAMIC_EXECUTION]: The Python files identified in the static analysis (scripts/tests/test_omni_contracts.py and scripts/tests/_bootstrap.py) are part of an internal test suite. The use of subprocess.run in these files is for automated unit testing of the skill's bash scripts in an org-free, network-free environment and does not pose a risk to the runtime execution of the skill.

service-omni-channel-setup-coordinate