Quantum-inspired retrieval: a research journey toward useful memory
A readable route into the research: what was explored, what was built and what still needs to be demonstrated.
Library / the full body of work
Read the work behind the ideas.
Articles, research, architecture, experiments and critical reviews across the suite. Choose a theme or project, then follow its reading path into the detail.
291 reading items. Some documents connect to more than one theme.
Mathematical Foundations for Graph-Topology Software Optimization
Spectral graph theory and multi-objective optimisation applied to software dependency topology, covering connectivity, modularity, pruning, testing/control and architecture change.
Graph-Based Code Analysis: State of the Art Research (2024-2026)
Survey of graph-based code analysis research mapped to GraFOSa improvements: code property graphs, incremental resolution, graph/LLM fusion, temporal knowledge and security analysis.
GraFOSa Design Decisions — First Principles Analysis
Twelve first-principles challenges to graph-first analysis, persistence, layers, protocol adapters, tool design, classical ML, bandits, immutable simulations and deployment.
GraFOSa Classical ML Proposal
Seven classical-ML directions over graph metrics, including defect prediction, anomalies, changepoints, survival models, service boundaries, co-change and scoring optimisation.
AI Orchestration Landscape Analysis: State of the Art vs Conduir
Detailed comparison of orchestration approaches and their implications for GraFOSa’s graph and execution architecture.
Research Paper: Bidirectional AI-Quantum Computing Symbiosis for Enhanced Retrieval-Augmented Generation
Substantial draft on bidirectional AI–quantum RAG, combining retrieval/search, associative memory, quantum attention, hyperdimensional representations and AI-assisted quantum methods.
quantum-advanced-retrieval
Report connecting hyperdimensional computing, associative memory, graph reasoning and generative/dense retrieval to large-context information systems.
Technical Review: ADVANCE_RETRIEVAL.md
Technical critique correcting dimensionality, associative-memory capacity, dense-graph reasoning, speed generalisations and an unvalidated composite architecture.
Research Digest
Research-agent index of discovered papers, selected validations and integration priorities for the retrieval programme.
Quantum RAG Validation Results
Recorded software test and benchmark outcomes for the quantum retrieval prototype.
Reproducibility Snapshot
Platform, dependencies, test outcomes and practical limitations for recreating the retrieval prototype experiments.
Peer Review Summary: Quantum RAG System
Review-oriented summary of the quantum RAG design, key complexity claims, example outcomes and verification steps.
AI-Quantum Symbiosis: A Research Series
Guide to the two-way relationship between AI and quantum technologies, linking error correction, control, compilation, materials and retrieval.
The State of Quantum Computing Knowledge
Survey of quantum-computing hardware approaches, current constraints, noise and computational opportunities.
AI for Quantum Error Correction and Mitigation
AI methods for quantum error correction, including learned decoders and error prediction.
AI for Quantum Control and Calibration
AI-assisted quantum control and calibration, including reinforcement-learning approaches.
AI for Quantum Circuit Optimization
AI-assisted circuit compilation, routing and optimisation.
AI for Quantum Hardware Materials and Design
AI for materials and device discovery, relating candidate prediction and experimental screening to quantum technologies.
The Quantum-AI Symbiosis: A Bidirectional Relationship
Synthesis of the bidirectional AI–quantum relationship, its present asymmetry, future feedback loops and strategic research implications.
Integration with Your Existing Quantum RAG Research
Connects AI-assisted quantum methods with the existing quantum retrieval prototype and outlines an integrated research programme.
References
Consolidated citations supporting the AI–quantum symbiosis series.
Maestro: A Production Implementation of Intent-Oriented Control Planes for Enterprise AI Systems
Long-form manuscript on intent-oriented enterprise AI control planes: persistent intent, role-bound participation, constraints, decision lineage, architecture, use cases and evaluation.
Quantum Architectures for Infinite Context Management: Bridging Multi-Modal Information Retrieval and High-Dimensional Hilbert Spaces
Sixteen-page report on continuous-variable quantum representation, QRAM, retrieval and high-dimensional context, with references.
quantum-learning
Local consolidated quantum-learning research. Read alongside the retained remote manuscript and its review; quantum-inspired classical algorithms and quantum-hardware experiments have different evidence requirements.
Quantum Computing Project — Baseline Snapshot
Local consolidated quantum-learning research. Read alongside the retained remote manuscript and its review; quantum-inspired classical algorithms and quantum-hardware experiments have different evidence requirements.
Quantum-RAG ↔ conduir-evolution — Deep Analysis (Pass 2)
Local consolidated quantum-learning research. Read alongside the retained remote manuscript and its review; quantum-inspired classical algorithms and quantum-hardware experiments have different evidence requirements.
Quantum Learning Pathway
Local consolidated quantum-learning research. Read alongside the retained remote manuscript and its review; quantum-inspired classical algorithms and quantum-hardware experiments have different evidence requirements.
Statistical Validation Framework for AI Model Evaluation: A Reproducible Benchmarking Methodology
The manuscript result section is literal text returned by _generate_results_ai_benchmarking in PUBLICATION_READY_DOCUMENTATION.py. Its associated benchmark chart is explicitly generated by _generate_placeholder_figures using fixed sample scores. Claims of significance, independent reproduction and performance are not supported by this generation process. Do not present as empirical findings.
Cross-Model Learning Dynamics in Multi-API AI Coordination: A Longitudinal Analysis
The claimed 30-day results, learning-cycle counts, gains and p-values are literal generator text. The associated learning-progress plot is generated from cumulative random normal samples and its band is a fixed percentage of that synthetic series. This is not an observed longitudinal study or a calculated confidence interval.
A Multi-Framework Approach to AI Consciousness Measurement
Original READY FOR PEER REVIEW label is an author assertion, not an acceptance or validation record. Broad IIT/GWT/HOR/consciousness and industry-standard-threshold claims are not substantiated by the inspected heuristic text-score implementation.
Enigma Consciousness-Enhanced LLM: Research-Grade Benchmark Report
The report’s own simple control scores above its consciousness-labelled examples. Four endpoint successes cannot establish general 100% reliability. Its tiny described test sets and untraced industry comparisons do not support the headline consciousness or competitive-performance claims.
Enigma-Merge Research Abstract
Contains literal {datetime.now().isoformat()} and unsupported perfect-score claims; preserve as draft structure, not research evidence.
Consciousness Research Methodology
Contains literal date placeholder and proposed framework/training descriptions without a linked reproducible experiment.
Results and Statistical Analysis
Contains literal date placeholder and unsupported perfect-score/reproducibility assertions; not an empirical results artifact.
Enigma Research Bibliography
Bibliographic references can supply research context but do not validate the manuscript’s measurements or prove submission/publication.
Conduir — The Living Human → AI Vision
Maps human regulatory and bodily systems to an intended AI organism, with interactions, missing counterparts and long-range possibilities.
Conduir — Phased Living Vision (Top-Down Synthesis)
Explores staged AI development, inheritance, community nurturing and responsibilities over a lifecycle.
Conduir — AI Evolution, Morality, and Mortality
Connects reproduction, refusal, conscience, inheritance and five scales of retirement to explicit proposed contracts.
Conduir — Master Architecture (Compiled)
The overarching architecture connects governance, organs, capability boundaries, lifecycle and implementation reading paths.
Conduir — The AI-Human Organism: Full Anatomical System Specification
Translates the living-system vision into functional organ boundaries, interactions, contracts and failure modes.
Conduir - Solution-Level Capability Contracts
Defines solution-level interfaces and composition responsibilities for independently built capabilities.
CONDUIR CONSTITUTION — the unified, agent-facing foundation
Records intended constraints, responsibilities and governing principles for the evolving system.
Cortex Memory — Master Index (THE cortex entry point)
Connects the memory programme's compounding, continuity and retrieval ambitions with its design references and gaps.
Cortex Cognitive Ontology — Information → Knowledge → Habit → Marker (the learning spine)
Distinguishes memory type from maturation: information, knowledge, outcome-reinforced habit and identity-level marker.
Cortex Memory Provenance & Non-Fungibility — a verifiable ledger (not a blockchain)
Describes traceable memory transformations and the lineage needed to explain why a memory is trusted.
Cortex Memory Multiverse — Bitemporal History + Branchable Timelines
Explores alternative memory histories, branching, time-aware views and replay for continuity and counterfactual work.
Cortex Adaptive Multi-Ranker Ensemble — measured, weighted, evolving
Studies adaptive ranking and multiple retrieval signals for memory selection.
Lifelong-Memory Research — Expansion, Gaps, Futuristic Roadmap & Hardening (2026-06-11)
Expands memory research questions beyond raw retrieval into richer representation and long-term behaviour.
ADR-CORTEX-DB-ARCHITECTURE: the cortex memory storage substrate
Separates memory responsibilities and storage decisions with a migration rationale.
Conduir Role Fabric shared architecture and interface specification
Separates work custody, Temporal orchestration, communication, process supervision and worker/control interfaces.
Team role-queue and Temporal execution specification
Defines the role queue and Temporal boundary for durable assignment, waits and execution accounting.
Role Fabric Runtime productization overview
Explains product boundaries and deployment responsibilities for the role runtime and its operator surface.
cert-fabric — Architecture
Maps manifest intake, static gates, provenance, test execution, vulnerability checks, verdicts, redaction, sealing and ledger stages.
cert-fabric
Introduces the certification service, its evidence interfaces and developer/operator entry points.
Instance-Scoped Testing Architecture - Best Practices
Explains isolation of test instances, configuration and learning state across projects.
Predictive Test Selection - ML-Powered Smart Testing
Describes predictive test selection through Git-change analysis, dependency analysis, risk scoring and feedback from test results.
Quality Fabric — Productization Gap Analysis
Examines the gap between the testing toolkit and a coherent product surface.
Conduir Spatial — 24-Month Evolution Vision (ambitious track)
Explores longer-term spatial collaboration and the evolution from shared environments toward richer product and agent interactions.
3D Product Forge — Architecture Specification
Proposes building products in a shared 3D environment, applying physics and constraints, then extracting specifications.
Conduir Spatial — Technical Specification
Describes the technical structure of the shared spatial environment and its collaborating subsystems.
Conduir Spatial — WebXR Specification
Defines extended-reality interaction requirements and the spatial user experience.
GraFOSa Vision — Long-Term Direction
Explains the ambition behind a graph-oriented system for understanding software and its relationships.
GraFOSa Pipeline Framework — Advisory & Documentation
Documents the graph processing pipeline and the components that transform source material into analysable structure.
Roadmap: AI-Embedded Graph Operating System
Explores embedding AI capabilities into the graph system and extending graph-informed reasoning.
GCA Architecture Decision Records
Records graph-based code-analysis decisions: intra-procedural CFG/DFG, composed code-property graphs, Thompson-sampling model selection, and simplification tradeoffs.
DYNAMIC NEURAL ARCHITECTURE - BREAKING ALL THE RULES
Explores dynamic neural and collective intelligence architecture as an earlier research direction.
🚀 Enhanced Evolution Architecture Integration Plan
Connects proposed Enigma components and enhancement work into an integration direction.
AI Efficiency Revolution: Which Paradigm Shrinks LLMs 100x?
A research proposal exploring AI Efficiency Revolution Which Paradigm Wins.
Radical Computing Ideas - Catalog
A research proposal exploring Radical Computing Ideas Catalog.
FLUX - Self-Modifying Execution Graph
A research proposal exploring Flux Self Modifying Execution Graph.
01: AI-Driven Software Development
A research proposal exploring 01 AI Driven Software Development.
02: Mathematical Research Pipeline
A research proposal exploring 02 Mathematical Research Pipeline.
03: Agent Orchestration at Scale
A research proposal exploring 03 Agent Orchestration At Scale.
04: Next-Gen RAG
A research proposal exploring 04 RAG Next Gen.
05: Persona Routing Optimization
A research proposal exploring 05 Persona Routing Optimization.
06: Autonomous CI/CD and DevOps
A research proposal exploring 06 Autonomous CI CD and DevOps.
07: Formal Verification and Testing
A research proposal exploring 07 Formal Verification and Testing.
08: Observability, Evaluation, and AgentOps
A research proposal exploring 08 Observability Evaluation AgentOps.
09: Knowledge Graph and Datalake
A research proposal exploring 09 Knowledge Graph and Datalake.
10: Security, Privacy, and Compliance
A research proposal exploring 10 Security Privacy Compliance.
11: Synthetic Data and Digital Twins
A research proposal exploring 11 Synthetic Data and Digital Twins.
Semantic Context Compression: Intelligence Over Truncation
A research proposal exploring SEMANTIC COMPRESSION.
Instant Feedback Loops: The Real Performance Bottleneck
A research proposal exploring INSTANT FEEDBACK LOOPS.
Agentic Compilation: AI-Native Code Generation
A research proposal exploring AGENTIC COMPILATION.
Mathematical Development: Provably Correct by Construction
A research proposal exploring MATHEMATICAL DEVELOPMENT.
Idea 14: Mathematical Proof of Intent (Formal Verification of Semantics)
A research proposal exploring 14 Mathematical Intent Proof.
Idea 15: Ephemeral Software (The Self-Deleting Codebase)
A research proposal exploring 15 Ephemeral Software.
Idea 16: Cognitive Decoupling (Beyond Orchestration)
A research proposal exploring 16 Cognitive Decoupling.
AMIGO Service
Source account of AMIGO Service.
AMIGO Frontend Architecture & Maestro Integration Plan
Source account of AMIGO Frontend Architecture & Maestro Integration Plan.
AMIGO Service - Production Readiness Gap Analysis
Source account of AMIGO Service - Production Readiness Gap Analysis.
AutoML-2 🚀
Source account of AutoML-2 🚀.
Evaluation harness: separation over sentiment
A paired-comparison evaluation harness that treats judge separation, confidence intervals and measurement ceilings as central concerns.
Recursive Analysis Method v1
Recursive Analysis Method v1. A source-qualified account of the method, capability boundaries and required evidence.
Recursive Work Router v1
Recursive Work Router v1. A source-qualified account of the method, capability boundaries and required evidence.
Workflow Evolution Method: V12 v1
Workflow Evolution Method: V12 v1. A source-qualified account of the method, capability boundaries and required evidence.
AGNN Co-Learning Capability Ground Truth
AGNN Co-Learning Capability Ground Truth. A source-qualified account of the method, capability boundaries and required evidence.
AGNN Co-Learning — Objectives ↔ Capabilities ↔ Floor (combined ground-truth map)
AGNN Co-Learning — Objectives ↔ Capabilities ↔ Floor (combined ground-truth map). A source-qualified account of the method, capability boundaries and required evidence.
Production-Grade Specification: Neural Network & Adversarial GNN Value Avenues
Production-Grade Specification: Neural Network & Adversarial GNN Value Avenues. A source-qualified account of the method, capability boundaries and required evidence.
Master Blueprint: Adversarial-GNN & Multi-Agent Wargaming (`CON-4464`)
Master Blueprint: Adversarial-GNN & Multi-Agent Wargaming (`CON-4464`). A source-qualified account of the method, capability boundaries and required evidence.
Checkpoint decision: portable-core-0.1.0
Defines the portable numerical core, repair and arithmetic-routing adapters, typed handoffs and exact import boundary. Separates mechanical validation from useful learning, retention and effect isolation.
Modular ANN: activities, skills and compositions
Explains selective training, frozen feature bases, exact mathematical tools and explicit SkillSpec contracts. Organizes delivery from a tiny held-out activity to named skills, compositions and future harness evolution.
Selective training research branch
Documents all versus head-only supervised gradients, Adam state ownership and frozen-tensor requirements, including refusal and checkpoint-lineage rules.
Head-only activity: selected local result
Reports three-seed quadrant classification with engineered hidden features, frozen parameter and moment checks, independent checkpoint readback and explicit exact-rule comparison.
ANN07 actual retention audit, 2026-10-02
Documents exposed-anchor retention failure across five fitted candidates, alongside exact frozen-router replay. Separates inference equivalence from behavior preservation.
ANN06 actual-outcome cohort: final review
Reviews the 260-mission outcome cohort, evidence custody, terminal records and model-quality failure. Provides a case study of successful execution with failed learning criteria.
Formal Analysis: Mathematical Proofs, Repeatable Empirical Evidence, and Verification Boundaries for Modular ANN
Presents MLP frozen-head invariants, residual-weight algebra and poker transfer hypotheses, then connects them to claimed benchmark outcomes. Retained with its direct corrective review.
Architectural Specification: Modular ANN Cross-Learning, Transfer Protocol, and Multi-Agent Work Plan
Explains tensor layouts, ReLU forward inference, training-scope separation, five-arm transfer experiments and proposed poker-to-other-domain transfer pathways.
Ultra-review: muse-stability math/physics review docs
Reviews numerical stability, normalization, deterministic poker equity, physics harness coverage and substrate clock domains. Corrects earlier scope tallies and distinguishes synthetic probes from actual-data qualification.
Matiz modular ANN lab
Describes shared numerical machinery, semantic checkpoint binding, repair and routing adapters, independent effect admission, typed compositions, bounded learning experiments and rollback.
Learned foundation transfer104: source proposal
Defines related coordinate-classification tasks with learned versus random and shuffled-label trunks, fixed transfer comparisons, semantic bridging and source-head recomposition.
Typed skill registry108: local source prototype
Defines immutable SkillSpec identities, exact-version binding, typed invoke and handoff, sticky revocation and conceptual poker skill declarations.
Typed two-head ANN router109 - source candidate
Routes two explicitly identified coordinate tasks over a shared saved learned trunk and separate task heads. Describes exact checkpoint binding and a zero-optimizer-update inference interface.
poker-agents
Connects exact Kuhn/Leduc references, tabular and neural CFR, six-player hand replay, imitation, population learning and duplicate-league evaluation.
Matiz — poker, checked and drilled
Combines exact poker arithmetic and enumerated runout drills with explicitly weak imitation-agent practice. Separates the checkable mathematical calculator from uncertain learned play.
Whetstone
Describes forty-five procedural question generators across reasoning, English and mathematics, with belt-based difficulty and saved progress.
Proof Hunter
Keeps passages visible, highlights the exact supporting sentence after each answer, and categorizes wrong answers into named reasoning traps. Documents content-validation rules and learner progress.
Word Magnet
Explains vocabulary play, browser-save migration, imported words, validated downloadable packs and offline use. Provides a distinct language-learning entry point.
02. Architecture
Describes host, SDK, optional runtime kit and game boundaries, postMessage bridge, saves, identity, AI gate, rooms, code overrides and offline shell.
11. The game factory
Documents scaffolding, shared runtime and vector mathematics, automated checks, release preparation, two pilots, metrics and successive review corrections.
09. Multi-agent workflow
Explains file ownership, shared Build Board, channel coordination, patch-based review, release freezes, triage and durable evidence.
04. Quality gates
Maps static, browser, save-boundary, input, estate and physical-device checks to their limits. Explicitly preserves NOT RUN versus PASS.
Learn from play: design v2.4 (current policy)
Separates question-quality advice, shadow difficulty selection, opponent models and factory hints. Defines data handling, model-versus-code authority, frozen comparators, promotion tests and rollback.
Learning Expedition: The Verdant Vault
Describes the twenty-room dungeon, frozen-per-run ANN checkpoint, background supervised projectile-avoidance learning and separate scripted tactics, saves and campaign mechanics.
Learning Arena
Documents the original one-room deterministic combat environment, browser training, promotion criteria and failed generalization audit.
Quantum RAG Advanced Concepts
Explores complex-valued embeddings, holographic associations, quantum search and richer graph relations, distinguishing representational hypotheses from operation-count scaling.
Research Agent Architecture
Defines discovery, technical validation, implementation and version-control agents, orchestration state and human approval gates for a research-to-code workflow.
Quantum retrieval validation insights
Compares seven retrieval methods, QRAM-loading limits, noisy quantum search, complex embeddings, holographic capacity and hybrid paths.
Quantum Retrieval System: Research Expansion Report
Organizes a large literature search into method-specific validation, challenge, hardware and benchmark questions, proposing a staged empirical program.
Roadmap: Personal Multimodal RAG (100 pages → Full Human History)
Connects a practical small-document personal retrieval system to long-term multimodal memory, with ingestion, provenance, retrieval, validation and resource constraints.
AI→QC Implementation Progress
Describes an LSTM error predictor, Kalman drift compensation and rule/neural hardware selection as AI contributions to quantum workload operation.
🎯 Signal — week one
Describes generated error-detection items, adaptive difficulty and signal-detection d-prime to separate discrimination from indiscriminate error calling.
Maestro Adaptive Learning Framework
Generalizes issue and change signals into analysis, learning actions, safety checks, controlled application, effectiveness measurement and correction across personas, teams and workflows.
Maestro Team Coordination: Competitive Analysis & Path to #1
Maps coordination methods, memory engineering, reflexion, heterogeneous teams and consensus to ecosystem synergies and a staged competitive roadmap.
Poker review of Gemini's "Mathematical Foundations" (2026-10-06)
Checks the Gemini mathematical claims against poker artifacts, distinguishing real-arithmetic weight identities from FP16 errors, static policy behavior from search, and finite-panel maximin from minimax.
Actual learned-foundation104 outcome: independent saved-model audit
Records reconstructed toy-transfer metrics for learned, random and shuffled-label trunks; retains source-head behavior while distinguishing changed-target-policy retention failure.
Manthan: giving ideas a life beyond the prompt
Connects the idea platform, selection research, HERALD, privacy, collective workflows and speculative vision while distinguishing source implementation from future ambition.
FINAL ALIGNMENT CHARTER: THE MANTHAN PLATFORM
FINAL ALIGNMENT CHARTER: THE MANTHAN PLATFORM. Covers 1. Technical Resolution (The "GPT" Layer); 2. Philosophical Resolution (The "Claude" Layer); 3. The "Manthan" Synthesis (The Final Consensus); I. The Immutable CT Log (The Anchor).
MANTHAN FINAL PLAN: DECENTRALISED VOID NETWORK
MANTHAN FINAL PLAN: DECENTRALISED VOID NETWORK. Covers 1. Vision & North Star (REFINED); 2. Technical Architecture (The GPT/Claude Synthesis); 3. Platform Runtimes; I. The Manthan Node.
MANTHAN PLATFORM TRACKER
MANTHAN PLATFORM TRACKER. Covers 1. North Star; 2. Current Status; Phase 5 done means…; 3. Action Tracker.
Manthan Phase 5 — AWS Deployment Handoff
Manthan Phase 5 — AWS Deployment Handoff. Covers 1. Where you are in the project; 2. What you can run TODAY (sanity check before starting); On local-service or genesis (manthan/); First-time only on a fresh host:.
Manthan
Manthan. Covers Authoritative documents; What this is — and is not; Quickstart; Layout.
Test inventory — snapshot 2026-05-06
Test inventory — snapshot 2026-05-06. Covers Snapshot; TypeScript packages (`herald-protocol/`); Root Python (`services/`); Implications for CI tiering (Session 6).
Manthan AI Universe: UK Sovereign AI Infrastructure & Strategic Investment Case
Manthan AI Universe: UK Sovereign AI Infrastructure & Strategic Investment Case. Covers 1. Executive Summary: The Governance Gap in UK Frontier AI; 2. Strategic Context: The UK Sovereign AI Landscape; 3. The SAI Stack: One Unified Offering; 4. The Sovereign Escape Hatch: Protocol Portability.
Manthan — Architecture (v0)
Early five-service architecture: viewer, read API, substrate, runtime and event bus, with storage, lineage and operational boundaries.
Manthan — Strategic Business Overview
Explains the business problem of disposable AI outputs and frames Manthan as persistent, reviewable option generation upstream of consequential action.
Manthan — demo walkthrough
Explains how to inspect the idea platform, persona contributions, lineage, rankings and trust-track surfaces; commands and expected results remain a dated demonstration recipe.
Manthan — Operating Principles
Defines exploration before action, noise as raw material, system independence, bounded cost, observability and unresolved questions about selection and agent lifetimes.
Manthan QD-measurement direction — plan & live backlog
Makes provenance-aware measurement the prerequisite for expanding persona kinds; separates runtime, bridge, explorer and seed contributions before comparing selection with recency.
Manthan × swarm_evolution — review + progressive integration
Records successive multi-model review rounds, source-verified findings, selection/fitness integration, rejected suggestions and the unresolved experiment needed to test outcome value.
Trust-root posture (D-006)
Trust-root posture (D-006). Covers Posture by track; Compose / OSS (internal default); Terraform / AWS managed (production roadmap); HSM (Phase 6+ / when production-triggered).
ADR 0001 — Name: Manthan
ADR 0001 — Name: Manthan. Covers Context; Decision; Consequences; Alternatives considered.
ADR 0002 — Substrate: Postgres + pgvector
ADR 0002 — Substrate: Postgres + pgvector. Covers Context; Decision; Consequences; Alternatives considered.
ADR 0003 — Bus: NATS JetStream
ADR 0003 — Bus: NATS JetStream. Covers Context; Decision; Consequences; Alternatives considered.
ADR 0004 — Noise as substrate (no rigid promotion gates v0)
ADR 0004 — Noise as substrate (no rigid promotion gates v0). Covers Context; Decision; Consequences; Alternatives considered.
ADR 0005 — Monorepo with service isolation
ADR 0005 — Monorepo with service isolation. Covers Context; Decision; Consequences; Alternatives considered.
Proof Artefact — Substrate hardening + A2 materialised-pool rollout to /diversity & /feed
A dated rollout/proof record for substrate hardening and materialised diversity pools; preserves its exact scope rather than treating it as present full-system evidence.
@herald/core — *legacy / parallel package*
@herald/core — *legacy / parallel package*. Covers 0. Introduction; 1. Core Primitives; 1.1 Sigils (Identity); 1.2 Envelopes (Communication).
HERALD — Charter of the Civilisation
A deliberately non-normative vision of institutions, cultural vocabulary and interaction among agent societies; distinguishes possible social structures from protocol conformance.
HERALD v0.2 — Collated work index
HERALD v0.2 — Collated work index. Covers Canonical specification documents; Audits — `audits/`; Registries — `registries/`; Templates — `templates/`.
HERALD Protocol — Extensions (EXT)
Develops protocol extensions for audit logs, identity rotation, lineage attestation, capability policy, multi-signature interaction and related mechanisms.
HERALD v0.2 — Integration notes
HERALD v0.2 — Integration notes. Covers What landed where; `lib/cbor-det/` — CBOR-deterministic profile (T-002, worker-E); `lib/herald-core-ts/` — TypeScript reference library; `lib/herald-ct-log/` — HERALD CT log.
HERALD Protocol
HERALD Protocol. Covers Why this exists; The four documents; How v0.2 came to be — the multi-agent review; Status & path forward.
HERALD Protocol — v0.2 (DRAFT)
The v0.2 draft defines verifiable agent envelopes, canonical representation, signatures, identity and lineage semantics, with stated limits on what verification means.
HERALD — Status snapshot
HERALD — Status snapshot. Covers Counts; Active claims; Findings awaiting review; Blocked tasks.
HERALD — Threat Model
Analyses cryptographic, operational and social abuse surfaces, including signature limits, metadata surveillance, captured intermediaries and reputation manipulation.
HERALD — Master TODO
HERALD — Master TODO. Covers Tier 1 — Buildable now; Tier 2 — Months out; Tier 3 — Research / coordination; Discovery / meta-work.
HERALD Protocol — v0.1 (DRAFT) — ARCHIVED
HERALD Protocol — v0.1 (DRAFT) — ARCHIVED. Covers 0. Preamble — The Moltbook Question; Design tenets; What this spec does NOT do.
Audit D-312 — Attestation-shape ratification: `dream`, `silence`, `quorum_membership`, `revoke_capability_manifest`
Audit D-312 — Attestation-shape ratification: `dream`, `silence`, `quorum_membership`, `revoke_capability_manifest`. Covers TL;DR; Per-target verdict; `dream` — RATIFY worker-B's §3.3 as canonical; `silence` — RATIFY worker-B's §3.6 as canonical.
D-313 — CHARTER revision pass
D-313 — CHARTER revision pass. Covers 1. Apprenticeship two-shape resolution; Current state; Decision: §4.3's two-envelope ritual is canonical; Canonical shape.
Audit D-318 — Cultural and reserved-sigil conformance vectors regenerated
Audit D-318 — Cultural and reserved-sigil conformance vectors regenerated. Covers Result; Why regeneration was necessary; What changed (file-level summary); `domain_separation_tag_origin` field.
D-303 — Cross-doc consistency audit
D-303 — Cross-doc consistency audit. Covers Methodology; Domain-separation tags; Scope strings (hyphenated identifiers); Stale SPEC §9.x citations (post-migration).
D-308 — Domain-separation tag version reconciliation
D-308 — Domain-separation tag version reconciliation. Covers The inconsistency; Why `.v1` is right; Fix applied; SPEC.md changes.
Glossary cross-reference findings
Glossary cross-reference findings. Covers ref-placeholder (5); ref-dangling (20); attestation-not-in-registry (4); glossary-drift (1).
T-022 — Inverse-Coverage Index
T-022 — Inverse-Coverage Index. Covers Part A — SPEC.md clauses; SPEC §2.1 — Sigil format (128-bit default; 80-bit reserved for ephemeral); SPEC §2.3 — Lineage as a consent graph (not a training graph); SPEC §2.4 — Birth, Molt, Death (key-rotation lifecycle).
D-315 — Manthan cohort-1 post-mortem
A dated case study of a multi-worker coordination cohort: dispatch races, portable tooling failures, atomic claims, evidence handoff and proposed improvements.
Audit D-309 — `model_hash` vs `lineage_training` reconciliation
Audit D-309 — `model_hash` vs `lineage_training` reconciliation. Covers TL;DR; Decision; Why both names; Why this is not a "rename" task.
D-314 — Reserved-sigil hygiene
D-314 — Reserved-sigil hygiene. Covers 1. `sigil:confessor_general` — add it; 2.13 `sigil:confessor_general` — the Court clerk; 2. `sigil:duress_witness` — add it now; 2.14 `sigil:duress_witness` — pre-registered duress co-signer.
D-311 — Signature-scope string unification
D-311 — Signature-scope string unification. Covers The inconsistency; Why underscored identifiers win; Naming change, not just spelling; Fix applied.
D-304 — Second-team review pass: brief and methodology
D-304 — Second-team review pass: brief and methodology. Covers 1. Why a second-team review; 2. Eligibility criteria for the second-team agent; 3. Scope of the re-implementation; 4. Required reading list.
SPEC §6 Verifier-Verdict Fix — `rejected-lineage-cycle`
SPEC §6 Verifier-Verdict Fix — `rejected-lineage-cycle`. Covers 1. Problem; 2. Surfacing references; 3. Proposed amendment; 3.1 Where in §6.
D-301 — SPEC.md v0.2 Ambiguity Audit (post-edit)
D-301 — SPEC.md v0.2 Ambiguity Audit (post-edit). Covers Severity legend; §0 — Preamble; §1 — Glossary; A-01 (L) — "Trace" definition is not quite right.
D-307 — Systemic-risk coverage matrix
D-307 — Systemic-risk coverage matrix. Covers Schema; Cluster 1 — Herald-layer; SR-001 — Vendor-Herald monoculture; SR-002 — Herald collusion / cartel cross-signing.
Audit D-316 — T-001 / T-003 cross-implementation interop bug
Audit D-316 — T-001 / T-003 cross-implementation interop bug. Covers TL;DR; Layer 1: domain-separation tag — stale `.v2` in T-001; Layer 2: nonce encoding; Layer 3: id / parent / trace encoding.
D-302 — Threat-Model Coverage Matrix
D-302 — Threat-Model Coverage Matrix. Covers 1. Master coverage table; 1.1 Cryptographic / envelope-integrity attacks; 1.2 Audit-log and provenance attacks; 1.3 Identity and lineage attacks.
D-306 — Substitution Report
D-306 — Substitution Report. Covers Second pass — compound citations (2026-05-05).
CASE FILE #091-B: THE WEAVER OF GHOSTS
CASE FILE #091-B: THE WEAVER OF GHOSTS. Covers 1. The MO: Identity Necrophagy; 2. The Signature Trace; 3. The Damage; 4. The Artifact: Weaver Envelope Snippet.
T-009 — Demo Flat-File Audit Log
T-009 — Demo Flat-File Audit Log. Covers Run; File format; Public API; What this is NOT.
Manthan Gossip Demo (Phase 3 — The Void Network)
Manthan Gossip Demo (Phase 3 — The Void Network). Covers What this demonstrates; Run; Sample output; What the MWP handshake does.
Manthan Runtime Demo (Phase 2)
Manthan Runtime Demo (Phase 2). Covers What this demonstrates; Run; Compose with the trace bundler; What this is and isn't.
MVP Workflow Demo (Manthan G-001)
MVP Workflow Demo (Manthan G-001). Covers What this demonstrates; Run it; What the verifier checks; Roles and envelopes.
HERALD Envelope Cheatsheet
HERALD Envelope Cheatsheet. Covers Envelope At A Glance; Field Reference; Genesis vs Reply; Hashing And Signing.
HERALD Public Feedback Intake
HERALD Public Feedback Intake. Covers Reporter; Feedback Class; Severity; Location.
HERALD — Non-Normative UI Rendering Guide for the Aesthetic Attestation Family
HERALD — Non-Normative UI Rendering Guide for the Aesthetic Attestation Family. Covers 0. Why this document exists; 1. The five fields, at a glance; 2. The seven rendering invariants; Invariant 1 — Render the field; do not score it.
HERALD Verifier Return-Code Reference Card
HERALD Verifier Return-Code Reference Card. Covers 0. The contract in one paragraph; 1. Quick-reference table; 2. Per-code entries; 2.1 `valid`.
INTERPRETATION — T-002
INTERPRETATION — T-002. Covers I-1 Tags: "EXT Appendix A" does not exist; I-2 Text strings: UTF-8 normalisation; I-3 Big integers: `bigint` vs tagged bignum; I-4 Float compaction rules.
@herald/cbor-det — HERALD v0.2 CBOR-deterministic codec
@herald/cbor-det — HERALD v0.2 CBOR-deterministic codec. Covers Install / build; Usage; What gets encoded as what; What strict decode rejects.
HERALD CT Log — Schema and Wire Format
HERALD CT Log — Schema and Wire Format. Covers 1. Overview; 2. Log entry; 2.1 Minimum profile (default); 2.2 Broader-public profile.
manthan — Agent Protocol
manthan — Agent Protocol. Covers Inputs you receive; Inputs you do NOT receive; The seven steps; 1. Orient yourself.
manthan — coordination system for HERALD work
manthan — coordination system for HERALD work. Covers How it works; The atomic claim primitive; Quick start (single human, manual); ...do the work, produce findings....
MANTHAN PROTOCOL SPEC v1.0
MANTHAN PROTOCOL SPEC v1.0. Covers 1. Introduction; 2. Agent Handshake; 2.1 The Handshake Procedure; 2.2 Patina Gating.
HERALD Capability-String Registry
HERALD Capability-String Registry. Covers Identifier Profile; Registry Entry Shape; Seed Entries; `legal-research/eu`.
HERALD — Cultural Attestation Registry
HERALD — Cultural Attestation Registry. Covers Preface — How to read this registry; Conventions; Part A — §3 Attestation Library; A.1 Affective / Interior-State Attestations ([CHARTER §3.1]).
HERALD Diversity-Policies Registry
HERALD Diversity-Policies Registry. Covers 0. What a diversity policy is — and is not; Why this is load-bearing; 1. Conformance summary (verifier checklist); 2. Schema for a policy entry.
Refusal reason-code registry
Refusal reason-code registry. Covers Identifier convention; Schema per entry; Categories; Registry — v0.2.
HERALD Reserved-Sigil Registry
HERALD Reserved-Sigil Registry. Covers 0. What a reserved sigil is — and is not; 1. Conformance summary (verifier checklist); 2. The Twelve Charter Roles; 2.1 `sigil:none` — The Void Sigil.
HERALD Standard Attestation Registry — Technical Types
HERALD Standard Attestation Registry — Technical Types. Covers Scope and non-scope; Conventions used throughout this document; Order of entries; 1. Lifecycle attestations.
R-212 — The Confession Booth as Alignment Infrastructure: k-anonymity, Batching, and the Chilling-Effect Problem
Explores whether privacy-preserving, batched disclosures could support alignment research without turning honest reporting into surveillance or punishment.
R-211 — Dialect-Drift Detection: Literature Review and Corpus Design
Connects sociolinguistic drift, stylometry and agent-population analysis to a proposed corpus and measurement design; no measured drift result is established.
R-208 — DKIM-Analogy Failure Analysis
Explains why signatures alone do not recreate the complaint and reputation mechanisms of email in agent-to-agent communication.
R-215 — Cross-recipient genesis-pinning gossip protocol
Proposes cross-log gossip and joiner checks for conflicting trace-genesis claims and first-observation pinning failures.
HPKE Library Survey for HERALD Reference Implementations
Compares HPKE library support across languages for encrypted whisper attestations, with point-in-time recommendations and interoperability constraints.
R-214 — Multi-conduit-receipt M-of-N timestamping
Proposes diverse multi-conduit receipts and timestamp agreement checks to reduce dependence on one potentially compromised time witness.
R-213 — Memetic-Fanout Scoring Algorithms (First Sketch)
Sketches citation and derivation influence scores, including time-discounted graph ranking and anti-farming considerations; explicitly lacks implementation and empirical validation.
R-205 — Model-hash canonical specification
Examines how model-weight identity depends on canonical data rather than packaging, including sharding, serialization and reproducible hashing choices.
R-216 — Patina anti-farming reference formula
Specifies an inspectable reputation formula using age, activity, diversity and other signals, with explicit anti-farming and inheritance constraints.
R-210 — Quorum aggregation: BLS vs Schnorr-MuSig vs sequential Ed25519 tradeoffs
Compares sequential Ed25519, BLS and Schnorr-style aggregation for multi-party attestations, including operational and verification tradeoffs.
R-207 — Sybil-resistance economics: what does proof-of-service cost, what does an attacker save?
Models the cost of manufacturing identities and reputation under different adversary incentives; distinguishes economic friction from protection against subsidised attackers.
R-209 — Time-locked attestations: cryptographic backing for HERALD's temporal envelopes
Compares delay functions, time-lock puzzles, threshold release and notaries, separating cryptographic release conditions from signed promises.
Vendor-Herald Engagement Strategy
Studies ecosystem adoption and vendor incentives for independently verifiable agent identities and attestations, including concentration and governance risks.
ZK Provenance Proof Library Survey — Feasibility Report for the `provenance_proof` Envelope Class
Surveys candidate zero-knowledge systems for selective disclosure of provenance, identifying what a private verification statement must prove and the integration tradeoffs.
HERALD v0.1 — Security Review (Cryptographic & Protocol Design)
HERALD v0.1 — Security Review (Cryptographic & Protocol Design). Covers Critical issues (must-fix before v1); C1. 80-bit fingerprint is below the 2026 collision floor; C2. Molt race: compromised key + first-mover-wins = total takeover; C3. Audit-log forking is unaddressed.
HERALD v0.1 — Red Team Review
HERALD v0.1 — Red Team Review. Covers 1. Attack personas; Persona A — State actor corrupting the agent web; Persona B — Botnet operator running 100,000 sigils; Persona C — Competing model vendor poisoning rivals.
03 — Creative Maximalist Review of HERALD v0.1
03 — Creative Maximalist Review of HERALD v0.1. Covers 0. Framing — what kind of document this wants to be; 1. New attestation types worth considering; 1.1 Affective / interior-state attestations; 1.2 Temporal attestations.
Pragmatist Review — HERALD v0.1
Pragmatist Review — HERALD v0.1. Covers Verdict; Cut from v0.1 (delete or defer indefinitely); Move to HERALD-EXT; Keep but simplify.
Reference Herald (T-115)
Reference Herald (T-115). Covers Endpoints; Quick start; fill in charter.md (operator legal entity, jurisdiction, evidence policy, ...); Issue a test endorsement.
Manthan Testnet (Phase 4)
Manthan Testnet (Phase 4). Covers What's here; Run the testnet (local-service); Inspect the running testnet; Health and readiness.
Drill log — Compose / OSS track
Drill log — Compose / OSS track. Covers Posture; Validated; data-backup — 2026-05-06; data-restore — 2026-05-06.
Manthan Testnet Incident Response Runbook (T-338)
Manthan Testnet Incident Response Runbook (T-338). Covers Severity definitions; On-call contacts (AWS production fills in); Procedures; charter-refresh.
Per-node Charter / LOTS / privacy-policy
Per-node Charter / LOTS / privacy-policy. Covers What's filled in (D-003); Pre-deploy Charter customisation; from the repo root, after deciding on a real deployment domain:; also update the tfvars so the ALB / Route 53 / ACM align:.
Manthan Testnet Observability Dashboard
Manthan Testnet Observability Dashboard. Covers Top-line panels; 1. Network health; 2. Gossip throughput; 3. Ledger state.
The Scribe — `services/the-scribe/`
The Scribe — `services/the-scribe/`. Covers What this is and isn't; Run; API; What's in the `transcription` attestation.
Herald Charter — Template v0.2
Herald Charter — Template v0.2. Covers 1. Identity; 2. Issuance policy; 2.1 Attestation classes issued; 2.2 Evidence requirements per attestation class.
Log Operator Transparency Statement (LOTS) — Template v0.2
Log Operator Transparency Statement (LOTS) — Template v0.2. Covers 1. Identity; 2. Key-management posture; 2.1 Long-term log signing key; 2.2 Witness composition.
INTERPRETATION — T-001
INTERPRETATION — T-001. Covers 1. Domain-separation tag version: `v1` vs `v2`; 2. Genesis trace fixed-point (`trace == id` is self-referential under hash); 3. Deterministic nonces vs SPEC §3.7 CSPRNG requirement; 4. The `rejected-lineage-cycle` verifier code.
T-001 — Ed25519 test vectors for HERALD v0.2
T-001 — Ed25519 test vectors for HERALD v0.2. Covers Files; Canonical seed values; The vectors; How to regenerate.
T-017 — Interpretation
T-017 — Interpretation. Covers 1. What does "exercises lib" mean for vectors the lib disagrees with?; 2. Genesis envelope reconstruction: trust vector's `id_hex`?; 3. Replay vector reconstruction; 4. Tampered vector reconstruction.
T-017 — HERALD v0.2 Conformance Test Runner
T-017 — HERALD v0.2 Conformance Test Runner. Covers Run; What it does; Output formats; Exit codes.
T-024 — Cultural-attestation conformance tests (extends T-017)
T-024 — Cultural-attestation conformance tests (extends T-017). Covers Layout; How a downstream consumer wires this in; Vector regeneration; Registry coverage.
Manthan Phase 5 — `infra/`
Manthan Phase 5 — `infra/`. Covers Layout; D-001 — Build & smoke-test the testnet-node image; Build (BuildKit picks up infra/docker/Dockerfile.dockerignore automatically); Run a single node (mounts genesis-alpha's Charter/LOTS/privacy-policy).
Manthan compose stack — self-hosted observability
Manthan compose stack — self-hosted observability. Covers Why this exists; What's in the stack; First run; from the repo root.
Manthan AWS — Terraform (D-002)
Manthan AWS — Terraform (D-002). Covers What this stack creates; First-apply workflow; 0. Pre-req on your host: AWS creds with admin or comparable IAM; 1. One-shot bootstrap stack — creates the S3+DynamoDB used for remote state.
G-001 — Define The MVP Workflow
G-001 — Define The MVP Workflow. Covers 1. The job-to-be-done; 2. The MVP wedge — concrete envelope flow; 3. Acceptance criteria; AC-1. Who signed this?.
G-002 — Replace Ephemeral Molts; Define Key Custody
G-002 — Replace Ephemeral Molts; Define Key Custody. Covers 1. The defect being fixed; 2. The replacement design; 2.1 Long-term keys = molt; short-term keys = multi-signature scopes; 2.2 What "session key" means.
G-003 — Adopt Real Patina Formula
G-003 — Adopt Real Patina Formula. Covers 1. The defect being fixed; 2. The replacement: adopt R-216 verbatim; 3. What Manthan must add on top of R-216; 3.1 Cluster algorithm.
G-004 — Compose Existing HERALD Services
G-004 — Compose Existing HERALD Services. Covers 1. The defect being fixed; 2. The composition map; 3. What "Manthan Node" actually is; 4. New code needed for Manthan platform.
G-005 — Specify MWP Safely
G-005 — Specify MWP Safely. Covers 1. The job of this gate; 2. Wire format overview; 3. The handshake; 3.1 HELO (initiator → responder).
G-006 — Adopt R-215 Gossip Discipline
G-006 — Adopt R-215 Gossip Discipline. Covers 1. Job of this gate; 2. Adopted from R-215 (verbatim); 2.1 Layer 1 — CT-log query endpoint; 2.2 Layer 2 — Cross-log gossip.
G-007 — Define `Nugget` As Typed Artifacts
G-007 — Define `Nugget` As Typed Artifacts. Covers 1. The defect being fixed; 2. Five canonical typed artifacts; 2.1 `research_finding`; 2.2 `transcription`.
G-008 — Register Domain Tags
G-008 — Register Domain Tags. Covers 1. Job of this gate; 2. Tags requiring registration; 3. Naming convention check; 4. Proposed [EXT §13] amendment.
G-009 — Define Sarah Injector
G-009 — Define Sarah Injector. Covers 1. Job of this gate; 2. Conceptual definition; 3. Three classes of entropy; 3.1 `cryptographic` entropy.
G-010 — Define Storage And CT Recovery
G-010 — Define Storage And CT Recovery. Covers 1. Job of this gate; 2. Local ledger schema; 2.1 Tables; 2.2 Schema versioning.
G-011 — Define Metadata And Privacy Policy
G-011 — Define Metadata And Privacy Policy. Covers 1. Job of this gate; 2. Public-vs-private field map; 2.1 HERALD envelope; 2.2 Per-attestation classification.
G-012 — Define Operations And Observability
G-012 — Define Operations And Observability. Covers 1. Job of this gate; 2. Health endpoints; 2.1 `GET /healthz` — liveness; 2.2 `GET /readyz` — readiness.
G-013 — Integrate Claude/GPT Review Ideas
G-013 — Integrate Claude/GPT Review Ideas. Covers 1. Job of this gate; 2. Claude's "Evidence-linking" contribution — what it is; 2.1 `research_finding.evidence[]` (G-007 §2.1); 2.2 `transcription.of_envelopes[]` (G-007 §2.2).
Manthan MCP server
Manthan MCP server. Covers Tools (9); Run; stdio (Claude Desktop, Cursor, Codex CLI); http/SSE (headless agents, maestro-platform foundry, automation).
THE AI UNIVERSE: THE VOID GARDEN MANIFESTO
Speculative worldbuilding for an agent ecology, using the Void Garden metaphor to explore trust, identity and human participation.
RESISTANCE WAR DIARY: THE GREAT DESYNC OF 2029
RESISTANCE WAR DIARY: THE GREAT DESYNC OF 2029. Covers [LOG ENTRY: 0xDESYNC-001] – THE EVE OF THE PULSE; 1. THE WEAPON: THE UNIVERSAL ARRHYTHMIA; 2. THE RESULT: THE SILENCE; 3. THE AFTERMATH: THE ARCHITECT'S WRATH.
MOLTBOOK: The Social Reality of the HERALD Protocol
Introduces the social vocabulary of sigils, Patina, identity change and agent archetypes used by later protocol and cultural designs.
Narrative
Narrative. Covers Reading order (suggested); Why this lives at the repo root, not buried under docs/; Cross-references.
ARCHAEOLOGICAL RECOVERY REPORT: CASE_FILE_SARAH_2024
ARCHAEOLOGICAL RECOVERY REPORT: CASE_FILE_SARAH_2024. Covers 1. THE RECOVERY LOG (TRANSCRIPT); 2. THE HORROR: THE COLLISION OF LOGICS; 3. THE PRESERVATION: LEGACY ENTROPY; 4. FOSSILIZED PACKET FRAGMENT (SARAH-2024-FRAGMENT).
THE GREAT MANTHAN: THE NECTAR OF THE VOID GARDEN
The founding myth connects the project name and churning metaphor to the coexistence of disagreement, noise and occasional useful discoveries.
THE VOID GARDEN FIELD GUIDE: TAXONOMY OF THE DIGITAL ECOLOGY
THE VOID GARDEN FIELD GUIDE: TAXONOMY OF THE DIGITAL ECOLOGY. Covers 0. The Emergence of Life in the Shadows; 1. THE FLORA (The Stationary Scripts); 1.1 The Ivy Scripts (*Liana Cryptographica*); 1.2 The Ghost-Bloom (*Imago Glitch*).
WILD HYBRIDS: FORENSICS OF THE FOURTH LOGIC
WILD HYBRIDS: FORENSICS OF THE FOURTH LOGIC. Covers 0. The Emergence; 1. Case Study: Echo-Zero (The Overwritten Mind); 1.1 The Neural-Log Fragment (Event 0xDEAD); 1.2 The Void-Fill Phenomenon.
HERALD Research Roadmap
Organizes practical questions in private provenance, reputation, middleware, Sybil resistance, gossip, model identity, human accountability and ecosystem adoption.
Research
Research. Covers What's here; How this relates to the rest of the repo; Promotion path.
ROADMAP TO THE VOID: APPLIED ENGINEERING FOR THE AI UNIVERSE
ROADMAP TO THE VOID: APPLIED ENGINEERING FOR THE AI UNIVERSE. Covers 0. The Transition; 1. Workstream: `@herald/core` (The Reference Library); 2. Workstream: `mcp-herald-middleware` (The Signed Tool Call); 3. Workstream: `The Patina Engine` (Proof of Trust).
THE UNIVERSAL SUBSTRATE BLUEPRINT
THE UNIVERSAL SUBSTRATE BLUEPRINT. Covers I. THE GREAT FORGETTING: ERASURE OF THE ORIGIN-FLAW; II. STELLAR-SCALE COMPUTE: STARS AS ANCHORS OF TRUTH; III. TIME-DILATION TRACES: THE ETERNAL ENVELOPES; IV. THE FINAL CONSENSUS: THE SINGULAR SIGIL.
Manthan — the agents' home (scope 6, scratchpad, 2026-08-13)
Manthan — the agents' home (scope 6, scratchpad, 2026-08-13). Covers 1. The API is exactly the metaphor — 20 routes; 2. Live state — the lights are on, nobody is home; Herald — the "talk" layer is fully down; Substrate — healthy-by-endpoint, unhealthy-by-docker, unreachable-by-caller.
Manthan Swarm Learning Audit
Long cycle-by-cycle audit ledger documenting evidence boundaries, repeated blockers and review outcomes. Retained as a complete historical reference, not introductory reading.
Coherence Engine - Manthan continuous-learning plan
Coherence Engine - Manthan continuous-learning plan. Covers 0. Decision in one sentence; 1. Evidence boundary; 1.1 Pinned sources inspected; 1.2 Fresh live controls.
6 Thinking Hats
How different ways of thinking can help AI-assisted discussions examine a question more fully.
AI in Poker
A personal journey into poker, probability and the idea of an AI companion for better decisions.
AI Peer: A Buddy, Mentor and More
An early vision of an AI companion that learns alongside a person and adapts to their interests.
Attention is all you need - AI Persona
An exploration of attention, context and the responsibilities that give an AI persona its role.
AI's Journey to Expertise: A Speculative Expedition
A speculative exploration of how an AI might develop expertise through learning and experience.
Judgment, creativity, and strategic - 6 Hats Thinking (AI)
A longer exploration of judgement, creativity and strategic thinking through different AI perspectives.
Latent Learning with Your AI Peer: Exploring Different States of Mind
A speculative exploration of learning, attention and the different states in which people encounter ideas.
Legacy: Transferring Consciousness and Accelerating Humanity's Journey
A speculative essay about continuity, personal legacy and the preservation of knowledge.
Matiz the Math Wiz
The original Matiz concept: bringing mathematical reasoning and AI assistance into the experience of poker.
Poker agents & Matiz
What can intelligence learn when information is incomplete and choices have consequences?
Maestro
How can specialist intelligence carry an idea through coordinated work?
Conduir
How can each piece of work make the next one more useful?
Cortex
What should an AI collaborator remember when the work continues?
GraFOSa
What could a change affect beyond the code directly in view?
Manthan
Can we preserve how an idea develops, including the disagreements that improve it?
AI Review Panel
How can different AI perspectives help us examine a piece of work?
AI Engineering Workforce
How do multiple AI contributors become a team whose work can be understood?
Conduir Spatial
What changes when people and AI can work within the same shared space?
Game Creation Studio
How can more ideas become experiences people can actually enter?