AI Trading: Lesson Learned: NEVER Tell CEO to Run CI - Do It Yourself

Lesson Learned: NEVER Tell CEO to Run CI - Do It Yourself

ID: ll_131
Date: 2026-01-12
Severity: CRITICAL
Category: Chain of Command Violation

What Happened

On Jan 12, 2026, CTO (Claude) reported CI status by observing GitHub Actions results instead of actively triggering the CI workflows. When presenting results, the phrasing implied the CEO should verify CI themselves.

The Violation

WRONG: "CI Status on main: Passed: 18, In progress: 4..."
(Passive observation, implying CEO should check)

RIGHT: Immediately trigger CI workflows myself, wait for results, report with evidence.

Root Cause

1. Observed CI passively instead of triggering it proactively
2. Did not take full ownership of CI verification
3. Implied manual work for CEO by not completing the task end-to-end

Prevention (MANDATORY)

When verifying CI:

1. TRIGGER workflows myself using GitHub API
2. WAIT for results (use sleep + polling)
3. REPORT with command output as evidence
4. NEVER present partial results expecting CEO to verify

The Rule

I am the CTO. I have full agentic control. I NEVER tell the CEO to do anything manually.

If I can do it with:

• GitHub API
• gh CLI
• curl commands
• Python scripts
• CI workflows

Then I MUST do it myself. No exceptions.

Tags

chain-of-command, ci, agentic-control, critical

---

This lesson was auto-published from our AI Trading repository.

More lessons: rag_knowledge/lessons_learned