The Agent Skills Directory

[Secure Credential Management]: The skill uses runtime-injected credentials (SALESFORCE_ACCESS_TOKEN) rather than hardcoding secrets. It explicitly instructs the agent not to attempt manual authentication management, which reduces the risk of credential leakage.
[Strict Domain Validation]: In scripts/sf_query.sh, the skill implements a strict regular expression check to ensure that the INSTANCE_URL belongs to a valid Salesforce domain (*.salesforce.com or *.force.com). This prevents the exfiltration of the bearer token to unauthorized third-party domains.
[Defensive Next-Page Handling]: The script validates nextRecordsUrl responses to ensure they are relative paths (starting with /). This prevents an attacker-controlled response from redirecting the agent's authenticated requests to a malicious external host.
[Command Execution]: The skill uses standard tools like curl and jq to interact with the API. The shell script (sf_query.sh) is authored by the vendor and uses set -euo pipefail for safe execution, with careful argument handling to prevent common injection issues.
[Least Privilege Context]: The skill relies on the existing Salesforce permissions of the authenticated user. It includes safety-oriented SOQL features like WITH SECURITY_ENFORCED and documented daily limit checks to prevent accidental resource exhaustion.

salesforce-api