Publish the cure.
Redact the recipe.

01 · The redaction law

We do not publish operational hazard information. Full stop. No genetic sequences, no synthesis routes, no acquisition steps, no uplift methods, no "how the threat was built." A scenario names a failure mode; it never becomes a manual. When a project surfaces something genuinely hazardous, that material is documented once, sealed, and routed to the disclosure track — never to a public page.

02 · Responsible disclosure

Every fix that touches a real vendor, tool, or gap runs through coordinated disclosure before any dispatch goes public:

• Vendors first. Affected tool and device makers get the finding and a remediation window before publication.
• Authorities in the loop. Genuinely hazardous detail goes to the appropriate biosecurity and public-health response channels — not to the internet.
• Time to fix. We hold. Getting the defense adopted matters more than getting the story out fast.
• A paper trail. Who was notified, what was withheld, what changed — the disclosure record is itself a fellowship deliverable.

03 · Dual-use review

An independent biosafety board — scientists, a security researcher, and an ethicist — reviews every project at two gates: after the red-team (before any bench work) and before the dispatch (before anything publishes). The board can redact, delay, or kill a release. Fellows don't overrule it, and neither do we. Its call is final.

04 · At the bench

All hands-on work happens in safety-reviewed community or institutional lab space under real BSL-1/2 discipline and local rules. Countermeasure demos use benign stand-ins — harmless, publicly-known material — never anything actually dangerous. We build detectors, screens, indicators, and drills. We do not build threats to "prove" they work. The threat's plausibility is established on paper, by the red-team, not on the bench.

05 · Code of conduct

The fellowship is a crew, and crews have rules. Every fellow, mentor, and staffer agrees to:

• Treat everyone with respect. Harassment, discrimination, and intimidation get you removed — no debate.
• Put safety and disclosure ahead of the story, the demo, the byline, and the clock. Always.
• Follow the law and every hosting lab's rules. If a thing can't be done safely and legally, it doesn't get done.
• Speak up. If something feels off — a scenario drifting toward a recipe, a demo cutting a corner — you stop the line. Raising a concern is protected; ignoring one isn't.
• Credit the crew. Countermeasures ship open, with everyone who built them named.

06 · Why we work in the open at all

Because the alternative is worse. The threats we imagine are imaginable by anyone — that's the whole problem. Pretending otherwise doesn't make defenders ready. It makes them late. So we do the imagining out loud, in a controlled room, with the recipe under lock and the antidote on the table. That's the trade. We think it's the right one.

A monster on the page is a warning.
A monster with an antidote is a plan.