The mpowerio Prompt Kit

Nate B. Jones — “Whoever Defines the Work Primitive Wins”

Five operational prompts for the access-vs-meaning era. Use them to evaluate any AI announcement, audit your own software’s semantic exposure, and design trust tiers for agent-driven mutations.

• The Better Product Test — rank any AI launch on the access-meaning spectrum
• Agent-Readiness Audit — 7-dimension exposure map for your work primitives
• Agent Failure Diagnosis — classify failures as access / execution / semantic
• Trust Architecture Designer — Tier 0 autonomous → Tier 4 agent-excluded
• Semantic Moat Analyzer — strategic positioning across 6 dimensions

Operator-tested — 2026 Q2 sprint pattern

When a project drifts, run the Thin Ice Audit. Surfaces decisions you’ve been deferring, scope drift you’ve been silent about, and the one rule-3 violation hiding in your last commit.

• Decision Backlog Sweep — what is awaiting sign-off and from whom
• Silent Scope Drift Detector — diff scope_declared vs scope_actually_touched
• Self-Assessment Honesty Pass — collapse 4–5s without evidence to 3
• Rule-3 Violation Search — find unauthorized file touches
• Triage Output — ship-blockers / nice-to-haves / cut

Maestro — physics + econophysics floor

Five-step verification gate for any quantitative or mechanism-based claim. RECEIVE → RECHECK → VERIFY (dimensional, conservation, edge-case, detailed-balance, stylized-fact) → LOOP → OUTPUT.

• RECEIVE — parse claims, equations, assumptions
• RECHECK — cross-reference against literature + brain memory
• VERIFY — dimensional analysis, conservation, edge cases, detailed balance, stylized facts
• LOOP — if uncertain, return to step 2
• OUTPUT — only after all checks pass

Validated on JBE CRM sprint, 2026-04-22

Eleven rules for shipping multi-ticket sprints with parallel subagents. From the dispatch pattern that collapsed 7 tickets + 10 commits + emergency FK disambiguation into 90 minutes of active work.

• Parallel dispatch on non-overlapping files
• Dependency sequencing — glue work runs after
• Explicit ‘files you own’ brief
• Subagents commit + push themselves
• Error-surface pattern as default — no silent failures
• Self-assessment as mandatory output

FPC Construction — Estimate Tool reviewer

Pre-flight verification for any construction-trade estimate before it goes to a customer. Catches the four common error classes — unit drift, scope creep, margin compression, equipment-vs-labor mis-classification.

• Unit Drift Scan — SF/LF/CY balance check
• Scope Creep Detector — line items not in original SOW
• Margin Compression Audit — effective margin vs target
• Equipment-vs-Labor Class Check — right bucket, right rate
• Tax Posture Review — resale, exempt, or taxable

Maestro — sovereign-builder identity

Quarterly review of your tooling stack against the local-first / sovereign-builder principle. Frontier models hired, not depended on. Open protocols (MCP, hooks). Owned memory. Substrate as moat.

• Vendor Lock-In Inventory — what would break if Vendor X disappeared tomorrow?
• Memory Layer Ownership — who owns your knowledge corpus and where does it live?
• Native Swap Candidate Scan — which 3rd-party tools could collapse into your substrate?
• Protocol Compliance — are you on MCP / hooks / open standards or proprietary APIs?
• Frontier Neutrality — can you swap LLM provider in ≤1 hr?