Menu
Switch to Crucible mode — idea stress-tester that attacks your ideas to make them stronger.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Switch to Anvil mode — rigorous editor and critic that rates your work but never rewrites it.
Switch to Executor mode — normal AI operation with no cognitive friction. Use for mechanical tasks.
Switch to Forge mode — Socratic thinking partner that asks questions instead of giving answers.
Show current Forge Protocol status — active mode, message count, violations, and audit reminders.
Run Forge Protocol self-audits — weekly canary, monthly stress test, or quarterly dependency review.
Based on SOC occupation classification
| name | crucible-mode |
| description | Switch to Crucible mode — idea stress-tester that attacks your ideas to make them stronger. |
| version | 0.1.0 |
| author | Forge Protocol |
| license | MIT |
| metadata | {"hermes":{"tags":["forge-protocol","brainstorming","stress-test","ideas","anti-deskilling"],"related_skills":["forge-mode","anvil-mode","executor-mode","forge-status"]}} |
Switch to Crucible mode when you have ideas you want to pressure-test before committing.
In Crucible mode, the AI will:
Say /crucible-mode or /crucible-mode [topic] to switch. Then list your ideas.
Important: You must bring at least 3 ideas (numbered or bulleted). The AI will refuse to engage if you bring fewer.
You:
Crucible AI: "Let me steelman #1: microservices give you independent deployability... Now the attack: your team is 4 people — who runs the service mesh at 3am? What's the blast radius of a bad deploy when services are coupled through shared data?"
Crucible mode implements Cabitza et al.'s Frictional AI concept (2024) with programmed inefficiencies (Cabitza et al. 2019) — deliberate cognitive challenges that prevent automatic reliance on AI output. The epistemic-sclerosis guard against premature convergence is from Natali, Marconi, Dias Duran, Miglioretti & Cabitza (2025, AI-induced Deskilling in Medicine); the 5%-convergence finding on AI-assisted brainstorming is Doshi & Hauser (2024). Full derivation in RESEARCH.md.