modes-of-reasoning-project-analysis
Made by JeffEvidence-first project analysis using selected reasoning modes as bounded cognitive operators, followed by independent claim audit and decision-focused synthesis. Use when a project decision needs multi-perspective analysis or epistemological review, or when ordinary review is likely to miss important assumptions.
Works With
Quality and Utility come from canonical skill metadata. Community ratings appear separately as stars when subscriber reviews exist.
Subscribe to download
Public previews are available, but downloads, CLI sync, and premium content require an active subscription.
Sign In to Subscribe- Full skill content with deterministic versioning
- CLI sync + local search index
- Access to Jeffrey’s premium library
Save to your account
Saved skills live in your cloud list. Run jsm sync on each machine to pull them locally.
What's New with this Skill
3Updates, enhancements, and release notes for this skill.
Updated: modes-of-reasoning-project-analysis, debating personas replaced by audited claims
The old workflow obligated each reasoning persona to make a distinctive contribution, so agents manufactured whatever their role stereotypically produces: forced disagreements, invented defeaters, confidence by headcount. The rewrite turns each mode into a bounded cognitive operator with an explicit contract, atomizes every output into typed claims, and hands each claim to an independent auditor who never sees who wrote it. A truthful null result now beats a plausible invention.
- Every mode runs under an explicit contract (input, operation, valid output, invalid moves, abstain condition); NO_MATERIAL_INCREMENT is a successful checked result
- Outputs become typed atomic claims in MOR_CLAIMS.jsonl, each audited author-blind under a resolution contract matched to its claim type
- Agreement carries no evidentiary weight by itself; synthesis counts independent evidence paths, and five paraphrases of one grep count as one observation
- Disagreement must pass a nine-point alignment test before it is reported; unsupported positions are rejected rather than kept for balance
- Asymmetric long-shot ideas survive as bounded HYPOTHESIS records with explicit disconfirmers and cheap discriminating tests, not as dissenting votes
- New dependency-free validator scripts/validate_mor_run.py checks claim IDs, status ownership, audit provenance, author/auditor separation, and final snapshots
- Three new references: CLAIM_SCHEMA.md for paths, IDs and JSONL fields, EPISTEMIC_INTEGRITY_PROTOCOL.md for good-faith reasoning rules, and EVALUATION_PLAYBOOK.md for offline evaluation with regression gates
Analysis now starts from a neutral case file (MOR_CASE.md) that pins down the decision question, loss function, constraints, and pivotal unknowns before any mode runs. Mode outputs are atomized into typed claims in MOR_CLAIMS.jsonl, each resolved under a contract matched to its type: observations need source inspection or reproduction, causal claims need mechanisms and discriminating evidence, value claims need an attributable authority, and recommendations need audited supporting beliefs plus reversal conditions. A claim's author can never be its final auditor, and static grep is ruled in
Enhanced: modes-of-reasoning-project-analysis, readiness wait fixed for current NTM
The agent-readiness wait used a bare --timeout=120, which current NTM rejects with a missing-unit error. It now passes 120s, so the reasoning swarm spawns without a stall.
- ntm --robot-wait --condition=idle --timeout=120s replaces the bare integer the binary no longer accepts
Modes of Reasoning Project Analysis
Spawn 10 agents, each applying a different reasoning mode from an 80-mode taxonomy, to analyze a project from different angles. A lead agent triangulates their findings into a report with provenance-tracked consensus, resolved disagreements, per-mode contribution scores, and new-idea proposals.
- 80-mode reasoning taxonomy across 12 categories (Formal, Ampliative, Causal, Strategic, Meta, and more)
- 7 epistemological axes (ampliative vs non-ampliative, descriptive vs normative, truth vs adoption, and more); the 10 selected modes must span at least 3 of them
- Triangulation protocol: 3+ mode convergence = high confidence kernel findings, with explicit conflict resolution for disagreements
- 10 mode stacks for specific project types: architecture review, bug hunt, creative innovation, security audit, strategy, and more
- Lead agent operator cards: cognitive moves like Kill Thesis, Blind Spot Scan, and Cross-Pollinate for rigorous synthesis
- Works beyond code: research papers, business plans, organizational processes, policy documents
This skill analyzes a project through ten reasoning lenses at once. Borrowing from NTM's ensemble pipeline, it assigns each agent in a Claude Code + Codex swarm a distinct reasoning mode (deductive logic, adversarial attack, systems thinking, counterfactual analysis, Bayesian updating, and more) drawn from an 80-mode taxonomy spanning 12 categories and 7 epistemological axes.