Skip to main content

MCP Prompts Reference

Built-in agentic prompts for common SquireX workflows. Trigger these from your AI agent.

review-agentforce-security​

When to use: Before any Agentforce metadata deploy. Run a full security scan and get a prioritized remediation plan.

What the agent does:

  1. Calls scan_agentforce on the full project
  2. Groups violations by severity
  3. For each Critical/High violation, calls explain_violation and suggest_fix
  4. Returns a ranked remediation plan

Example prompt:

"Review my Agentforce metadata for security issues"

fix-agentforce-violation​

When to use: You have a specific SARIF violation and want a targeted fix.

What the agent does:

  1. Accepts a rule ID and file location
  2. Calls get_rule_details for remediation guidance
  3. Reads the violating file
  4. Calls suggest_fix with the relevant context
  5. Returns a diff-style fix

Example prompt:

"Fix the AGENTFORCE-1.1 violation in Submit_Case.genAiFunction-meta.xml"

harden-agent-metadata​

When to use: Proactive hardening review before going live with a new agent.

What the agent does:

  1. Runs scan_agentforce with all Critical and High rules
  2. Also runs instruction integrity rules (9.1, 9.2) with enhanced context
  3. Returns a defense-in-depth gap analysis

generate-test-evaluation​

When to use: You want to create Agentforce DX test specs for Testing Center.

What the agent does:

  1. Calls scan_agentforce to get violations
  2. Calls generate_dx_tests to convert to test YAML
  3. Optionally calls validate_dx_tests
  4. Returns the test spec YAML

Example prompt:

"Generate Testing Center test specs from my latest Agentforce scan"