Gemini/Codex Workflow Automation
Gemini CLI · Codex CLI · agentic workflows · durable context
What this proves
- Problem: long-running AI coding workflows break when context, handoff, and validation live only in chat memory.
- Method: built a file-driven orchestration environment around declarative intent, reconciliation, bridge scripts, MCP / n8n automation lanes, explicit guardrails, and deterministic close-out gates.
- Outcome: created a working automation system with
100scripts,34docs, and58research artifacts in the read-only source repo.
Why it matters for agentic workflow roles
This is the portfolio's clearest proof that Michael has worked beyond one-off AI prompting. The system treats AI-assisted work like an operating workflow: agents need durable context, defined handoff paths, validation gates, rollback-minded close-out, and public-safe documentation.
Recruiter-safe role framing:
- AI workflow specialist
- agentic automation specialist
- analytics automation engineer
- business systems automation analyst
- BI automation analyst
Public-safe proof surface
This portfolio only exposes the public-safe layer of the system:
- the documentation repo
- the architecture/runbook narrative
- the operating ideas that make the workflow durable
The larger working repo remains private and local-only.
Public proof themes
- durable context
- declarative intent
- reconciliation loop
- blackboard / swarm coordination
- Gemini/Codex bridge interoperability
- MCP / n8n automation lanes
- guardrails and tool boundaries
- deterministic close-out and validation gates