platform-soql-query

Pass

Audited by Gen Agent Trust Hub on Oct 8, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill leverages the Salesforce CLI (sf) to perform query plan analysis and execute queries, as documented in references/cli-commands.md. The validation hook script, scripts/post-tool-validate.py, interacts with the CLI to assess the relative cost and selectivity of generated queries, which is a core feature for query optimization.
  • [DYNAMIC_EXECUTION]: The skill provides numerous Apex templates and references, such as assets/selector-class.cls and references/selector-patterns.md, which demonstrate the use of dynamic SOQL via Database.query(). Additionally, it includes a Python script scripts/post-tool-validate.py that runs as a post-tool hook to perform static and live analysis of SOQL files.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes natural language input to generate SOQL and reads user-edited files for validation purposes. The attack surface is mitigated by the fact that the generated code is focused on read-only SOQL and SOSL queries, and the instructions consistently promote the use of security enforcement clauses like WITH SECURITY_ENFORCED and WITH USER_MODE.
  • [SAFE]: The skill is authored by the platform vendor (forcedotcom) and focuses on professional development workflows. All external tools (sf, jq, python3) and referenced technical blogs are standard resources within the Salesforce ecosystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 8, 2026, 05:31 AM