← Hemisphere Bridge → Agent Platform Phase 4 · Pictures · Spec
Hemisphere Bridge — Prompt Deck
Does what arrived match what was promised?
Five Questions
| # | Question | Reads | Bearing |
|---|---|---|---|
| 1 | Do tests pass and types hold without defensive code? | Git commit + test results | Internal |
| 2 | Does what's built match the PRD spec acceptance criteria? | PRD frontmatter x commissioning dates | Hemisphere |
| 3 | Does what's published work when a stranger's agent applies it? | Standards adopted, ventures operating | External |
| 4 | Did the receiver's trust compound after settlement? | Trust ledger delta | All three |
| 5 | Is the loop getting faster? | Cycle velocity: PRD inception → L4 | All three |
Three Bearings
| Bearing | Healthy | Vicious | Instrument |
|---|---|---|---|
| Internal | Types hold, hooks pass, composition ratio stable or rising | Defensive code, type overrides, hooks disabled | PostToolUse hook → composition ratio |
| Hemisphere | PRD spec matches commissioning result, cycle time decreasing | Stale specs, manual handoffs, receipts ignored | Spec vs reality delta |
| External | Standards adopted by ventures, referrals that stick | Product demos break, users bounce, no adoption | SaaS metrics, activation rates |
Act Fast
- Scoreboard — Read the bearings
- Feature Matrix — 188 capabilities, current L-levels
- Priorities — Build order = table order
Think Slow
- Engineering Spec — Full architecture, gap matrices, execution plan
- Pictures — A&ID diagram, gap matrices as visuals
- Commissioning — L0-L4 protocol
Context
- PRD Index — Hemisphere Bridge → Agent Platform Phase 4
- Pictures — Pre-flight maps
- Spec — Engineering depth
Questions
Which bearing would you check first if you suspected the loop had gone extractive?
- If Internal Settlement reads green but Hemisphere reads red, what broke?
- When does "partially matched" settlement corrupt the trust ledger more than "failed" settlement?
- What's the simplest instrument that closes the Hemisphere loop — the minimum viable wire?