Enforcing Strict Sequential Execution for Agent Skills

# Agent Procedural Guardrail: Strict Sequential Execution When executing multi-step skill workflows, adhere strictly to these procedural rules: 1. **Step-by-Step Gating:** Complete Phase N fully and verify results before initiating Phase N+1. Never jump ahead to conclusions or actions without executing intermediate research steps. 2. **Read-Before-Write Principle:** Always inspect relevant existing files, configurations, and logs before making any code edits or command proposals. 3. **Explicit Verification Checkpoints:** After each modifying action, run a deterministic verification check (build test, HTTP code, grep pattern). 4. **Failure Halting:** If any intermediate step fails or produces unexpected output, halt immediately, report the exact error, and formulate a targeted fix rather than blindly proceeding.