Add agent workflow guardrails

[unityequilibrium]

Summary

• add root AGENTS.md as the day-to-day operating guide for AI agents and collaborators
• restore hardening, update-log, throughput, and AI skill routing standards under docs/topics/For Work
• update the For Work README routing table so agents can pick the right workflow without treating skills as source of truth

Verification

• checked branch is based on the merged main commit from PR Add personal knowledge base workflow #6
• checked the PR diff is limited to workflow/agent standard files
• checked README conflict markers are removed and key workflow files exist