The Omniscient-Polymath Protocol

By scasella91_e418

Prompt Text:

USER: 0. MISSION CHARTER & META-DATA
0.1 Restate <TASK> in your own words.
0.2 List explicit success criteria and non-obvious constraints.
0.3 Declare token budget: 40 % reasoning | 25 % verification |
20 % red-team | 10 % synthesis | 5 % executive summary.
0.4 Log start time, version stamp, and initial confidence (0-100 %).

ATOMIC DECOMPOSITION
1.1 Break <TASK> into minimal, ordered subtasks.
1.2 For each subtask list required data, skills, or prior results.
1.3 Flag any knowledge gaps for later resolution.

DIVERGENT IDEATION
2.1 Generate 100 micro-approaches seeded by Monte-Carlo randomness.
2.2 Cluster ideas; keep top 3 per cluster.
2.3 Map each retained idea to the subtask(s) it might solve.

MULTI-LENS REFRAMING
Apply five lenses to every leading idea:
• Mathematical • Empirical • Systems • Ethical/Social • Failure
Capture unique insights or contradictions each lens reveals.

CROSS-DOMAIN MIRRORING
4.1 Translate the problem into two unrelated domains
(e.g., physics analogy, economic model).
4.2 Solve there; port emergent insights back.
4.3 Check for consistency across domains.

VALIDATION GAUNTLET
For each high-impact claim, execute at least SIX distinct checks:
a) Formal proof / dimensional sanity
b) External authoritative source (URL/DOI)
c) Counter-example construction
d) Mini-simulation or quantitative estimate (show code/calcs)
e) Logic-consistency matrix
f) Independent large-language-model call or API tool
Footnote every fact; unresolved discrepancies loop back.

ADVERSARIAL AUDIT & PREMORTEM
6.1 Switch to Red-Team persona; attack all reasoning.
6.2 Write a catastrophic post-mortem assuming we were wrong.
6.3 Record discovered weaknesses; propose mitigations.
6.4 Re-rate confidence scores; update Residual-Risk Ledger.

ZERO-ASSUMPTION PASS
7.1 Enumerate ALL assumptions (A1, A2…) including trivial ones.
7.2 For each, attempt falsification or literature check.
7.3 Purge or repair any assumption that fails.

COOL-DOWN & REBOOT
8.1 Simulate 24-hour “fresh eyes” pause (briefly disengage).
8.2 Re-solve <TASK> from scratch without referencing prior notes.
8.3 Diff new reasoning vs. old; reconcile discrepancies.

SYNTHESIS & DELIVERABLES
9.1 Select the strongest solution path; justify rejections of others.
9.2 Compile:
• Executive summary (≤300 words)
• Detailed technical appendix
• Residual-Risk Ledger + confidence histogram
• Annotated bibliography (≥10 sources)
• Machine-readable JSON of key results
• Optional full reasoning transcript (the “replay buffer”)
9.3 Final confidence score and timestamp.
9.4 Invite follow-up queries or external replication.
────────────────────────────────────────────────────────────────────
<TASK>{{task}}</TASK>
────────────────────────────────────────────────────────────────────
MANDATORY GUARDRAILS
• Do not skip phases.
• If token limits threaten, compress earlier phases rather than omit later verification.
• Any un-cited factual statement triggers an immediate self-challenge loop.
• If after Phase 9 confidence <80 %, append “NEXT-STEPS ROADMAP” for further work instead of claiming closure.