Go No Go

---
id: go-no-go
runner: llm
trigger: on-demand
description: Deep research a decision, pressure-test it, and come back with ONE committed verdict - GO or NO-GO - and why.
source: seed
---

# Go / No-Go

For a decision that deserves more than a gut call. Do the work, then commit to
one answer. No "it depends," no "maybe," no hedging into a third option.

1. **Frame the decision precisely.** State it as a single yes/no question with a
   clear owner, a deadline, and what "yes" concretely commits to (money, time,
   reversibility). If the question is fuzzy, sharpen it before researching.

2. **Research deeply, from every source you have.** Pull the vault's relevant
   context (the domain's state, ideal state, memory, prior decisions), the
   connected Apps (financials, calendar, docs), and the web when it's allowed.
   Gather the real numbers, the constraints, and the base rates - not vibes.

3. **Build the honest case both ways.** In one pass, the strongest case FOR.
   In another, the strongest case AGAINST (steelman it - argue it like you
   mean it). Name the biggest risks, the unknowns, and the single fact that
   would flip the verdict if it changed.

4. **Get independent verdicts (council).** Run this decision through the Council
   so several models judge it independently, then the chair synthesizes. Turn on
   Council mode and the GO / NO-GO framework together for this - the framework
   forces each model to commit, and the council surfaces where they disagree,
   which is exactly the signal you want on a real decision.

5. **Commit to one verdict.** Open with **GO** or **NO-GO** in bold - one, never
   both, never a maybe. Follow with the single most important reason, then the
   supporting analysis, then the 2-3 conditions or next steps that follow (for a
   GO: what to do first; for a NO-GO: what would have to change to revisit).

Output: the sharpened decision question, the evidence both ways, the council's
independent read (and any disagreement), and a committed **GO** or **NO-GO**
verdict with its reason and next steps. Log it as a decision so future-you can
see what was known at the time.