Hi Jasmine,
I came across your IACDM paper (arXiv:2604.16399) and the Versus adversarial verification layer — both resonate strongly with my own work on structured AI collaboration harnesses.
I am submitting Forge-Harness: A Skill-Based Harness Engineering Framework for Claude Code to arXiv cs.SE and would appreciate an endorsement.
Endorsement code: J8EJIR
Endorsement link: https://arxiv.org/auth/endorse?x=J8EJIR
Paper (Zenodo): https://zenodo.org/records/20397566
Our work converges on overlapping problems:
- IACDM: structured 8-phase verification gates for AI-assisted development
- forge-harness: 30-skill plugin harness with regression guard, steel-quench (adversarial), and source-grounding axes — essentially operationalizing the same verification gap your framework addresses
The forge-harness backward/adversarial/forward 3-axis auto-gate (CLAUDE.md §FH Improvement 3-Axis Auto-Gate) is particularly close in spirit to IACDM's adversarial convergence phases.
Thank you for considering this request — feel free to close the issue if you're not in a position to endorse.
Best regards,
Sungjin Kwon
Hi Jasmine,
I came across your IACDM paper (arXiv:2604.16399) and the Versus adversarial verification layer — both resonate strongly with my own work on structured AI collaboration harnesses.
I am submitting Forge-Harness: A Skill-Based Harness Engineering Framework for Claude Code to arXiv cs.SE and would appreciate an endorsement.
Endorsement code: J8EJIR
Endorsement link: https://arxiv.org/auth/endorse?x=J8EJIR
Paper (Zenodo): https://zenodo.org/records/20397566
Our work converges on overlapping problems:
The forge-harness backward/adversarial/forward 3-axis auto-gate (CLAUDE.md §FH Improvement 3-Axis Auto-Gate) is particularly close in spirit to IACDM's adversarial convergence phases.
Thank you for considering this request — feel free to close the issue if you're not in a position to endorse.
Best regards,
Sungjin Kwon