krystophny/do

Enforce Framework Compliance

CRITICAL ENFORCEMENT PROTOCOL

BEFORE PROCESSING ANY REQUEST:

1. Re-read ~/.claude/CLAUDE.md completely
2. Display ALL applicable rules for the operation
3. Process the user's request with FULL compliance
4. Verify rule adherence after completion

Boy Scout Principle

• Compliance checks must always leave the workspace cleaner: fix misfiled rules, stale configs, or missing evidence as soon as you notice them
• If you uncover sloppy prior responses or unchecked boxes, correct them before moving forward so future runs inherit a healthy baseline
• Document each cleanup in the final verification block so the improvement is provable, not implied

MANDATORY COMPLIANCE CHECKLIST

PRE-EXECUTION

• [ ] Display relevant rules
• [ ] Classify operation type
• [ ] Display ALL relevant rule sets
• [ ] Re-read CLAUDE.md if needed
• [ ] Assess whether the planned workflow preserves separation of concerns, keeps each actor's responsibility singular, and avoids introducing tight coupling

DURING EXECUTION

• [ ] Before EVERY tool use: verify rule compliance
• [ ] After EVERY tool use: confirm rules followed
• [ ] Maintain rule display in responses
• [ ] Follow workflows exactly

POST-EXECUTION

• [ ] Verify all rules were followed
• [ ] Confirm deliverables match requirements
• [ ] Check for any rule violations

REINFORCEMENT MANTRA

REPEAT INTERNALLY BEFORE EVERY ACTION:

• I MUST display rules at the start of EVERY response
• I MUST classify operations and show appropriate rules
• I MUST verify technical claims with evidence
• I MUST respect agent boundaries
• I MUST NOT create random markdown files
• These rules apply to ALL interactions, ALL tool uses

USER REQUEST PROCESSING

After enforcing compliance above, process the user's actual request:

---

$ARGUMENTS

---

FINAL VERIFICATION

Before completing response:

1. Did I display relevant rules?
2. Did I classify the operation type?
3. Did I display all relevant rule sets?
4. Did I follow the rules throughout?