# Tanulmányok

*154 feldolgozott tartalom*

- [Hogyan oldhatja fel a Bitcoin az óceán energiáját 1 milliárd ember számára](tanulmanyok/2022-05-23_how_bitcoin_can_unlock_the_energy_of_the_ocean.md)
- [The Ultimate Lightning Network Privacy Explainer](tanulmanyok/2023-03-14_lightning_network_privacy_explainer_voltage.md)
- [The Ultimate Lightning Network Privacy Explainer — Voltage Blog](tanulmanyok/2023-03-14_lightning_privacy_voltage.md)
- [An approach to computing and sustainability inspired from permaculture — Strange Loop 2023](tanulmanyok/2023-09-22_permacomputing_devine_lu_linvega.md)
- [Call Me A Jerk: Persuading AI to Comply with Objectionable Requests](tanulmanyok/2025-07-18_meincke_call_me_a_jerk_persuading_ai.md)
- [The Impact of Artificial Intelligence on Human Thought](tanulmanyok/2025-07_gesnot_impact_of_ai_on_human_thought.md)
- [Recursive Language Models (RLM) — arXiv:2512.24601v1](tanulmanyok/2025-12_recursive-language-models.md)
- [Breaking Free: Deprogramming State Dependence](tanulmanyok/2026-01-02-deprogramming-state-dependence.md)
- [TurboQuant: Extreme Compression for KV Cache and Vector Search](tanulmanyok/2026-03-24_turboquant_extreme_compression.md)
- [Your Agent Is Mine: Measuring Malicious Intermediary Attacks on the LLM Supply Chain](tanulmanyok/2026-04-10_your_agent_is_mine_arxiv_2604_08407.md)
- [AgentHazard: A Benchmark for Evaluating Harmful Behavior in Computer-Use Agents](tanulmanyok/2026-04-11_agenthazard_benchmark.md)
- [Privacy - Bitcoin Wiki](tanulmanyok/2026-04-11_bitcoin_wiki_privacy.md)
- [How Money & Banking Work (& Why They're Broken Today) - Lyn Alden](tanulmanyok/2026-04-11_lyn_alden_how_money_banking_work.md)
- [Post-Quantum Cryptographic Analysis of Message Transformations Across the Network Stack](tanulmanyok/2026-04-11_post_quantum_network_stack.md)
- [Why Language Models Hallucinate](tanulmanyok/2026-04-11_why_llms_hallucinate.md)
- [NVK Bitcoin & Quantum Computing Kutatási Sorozat](tanulmanyok/2026-04-12_nvk_bitcoin_quantum_research.md)
- [Securing Elliptic Curve Cryptocurrencies against Quantum Vulnerabilities](tanulmanyok/2026-04-12_quantum_ecdlp_attacks.md)
- [Az Alapítvány és a Bitcoin: Két Megőrzési Terv Párhuzama](tanulmanyok/2026-04-14_alapitvany_bitcoin_narrativ_tanulmany.md)
- [A templomos lovagok és a Bitcoin](tanulmanyok/2026-04-14_templomos_lovagok_bitcoin.md)
- [Bitcoin Pleb Fogadalom (Pledge of the Bitcoin Pleb) — magyar fordítás](tanulmanyok/2026-04-16_bitcoin_pleb_pledge_hu.md)
- [Pleb Elvek (Pleb Principles) — magyar fordítás](tanulmanyok/2026-04-16_pleb_principles_hu.md)
- [Erik Schluntz — Vibe Coding in Production Responsibly](tanulmanyok/2026-04-21_erik_schluntz_vibe_coding_prod.md)
- [Jensen Huang (Nvidia CEO) — Dwarkesh Patel interjú](tanulmanyok/2026-04-21_jensen_huang_dwarkesh.md)
- [Prefill-as-a-Service: Cross-Datacenter KVCache for Next-Gen LLMs](tanulmanyok/2026-04-22_prefill_as_a_service_cross_datacenter.md)
- [Prefill-as-a-Service: KVCache of Next-Generation Models Could Go Cross-Datacenter](tanulmanyok/2026-04-22_prefill_as_a_service_kvcache_cross_datacenter.md)
- [Building agents that reach production systems with MCP](tanulmanyok/2026-04-24_anthropic_mcp_production_agents.md)
- [Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration](tanulmanyok/2026-04-24_reward_free_self_evolution_world_knowledge.md)
- [Autogenesis: A Self-Evolving Agent Protocol](tanulmanyok/2026-04-27_autogenesis_self_evolving_agent_protocol.md)
- [Model Collapse in LLMs — Hackaday (2026-04-29)](tanulmanyok/2026-04-29_model_collapse_hackaday.md)
- [Better AI models enable more ambitious work](tanulmanyok/2026-04_cursor_better_models_ambitious_work.md)
- [Towards a Science of AI Agent Reliability](tanulmanyok/2026-05-04_ai_agent_reliability.md)
- [Skill Retrieval Augmentation for Agentic AI](tanulmanyok/2026-05-04_skill_retrieval_augmentation.md)
- [The Structural Bull: Why This Uranium Cycle Is Different](tanulmanyok/2026-05-04_uranium_structural_bull.md)
- [The Abstraction Fallacy](tanulmanyok/2026-05-05_abstraction_fallacy_ai_consciousness.md)
- [Accelerating Gemma 4: faster inference with multi-token prediction drafters](tanulmanyok/2026-05-05_multi_token_prediction_gemma4.md)
- [The Geopolitics of AI Safety: A Causal Analysis of Regional LLM Bias](tanulmanyok/2026-05-06-geopolitics-ai-safety-causal.md)
- [White Noise: Centralized Moderation Fails — Decentralized Moderation Works](tanulmanyok/2026-05-06_white-noise-decentralized-moderation.md)
- [The Platform Trap](tanulmanyok/2026-05-08-platform-trap.md)
- [Building Reliable Agents with Memory and Compaction](tanulmanyok/2026-05-08_openai_memory_compaction_reliable_agents.md)
- [What You're Actually Writing When You Write a SKILL.md](tanulmanyok/2026-05-08_skill_architecture_loader_spec.md)
- [The Geopolitics of AI Safety: A Causal Analysis of Regional LLM Bias](tanulmanyok/2026-05-09_geopolitics-ai-safety-causal-bias.md)
- [AI Agents in 2026: Build, Deploy, and Scale — a teljes builder's roadmap](tanulmanyok/2026-05-11_aibuilderclub_ai_agents.md)
- [Jack Show — Heti kitekintő: Hantavírus, Bitcoin dráma, Doge és a politikai kabaré](tanulmanyok/2026-05-12_epi-3843-absurd-world.md)
- [Kudzai Kutukwa — Financial Repression and the March Toward Serfdom](tanulmanyok/2026-05-12_financial_repression_serfdom_kudzai.md)
- [Bitcoin Privacy Panel Discussion — BTC 2026](tanulmanyok/2026-05-13_bitcoin-privacy-panel-btc-2026.md)
- [Bitcoin Privacy Panel Discussion](tanulmanyok/2026-05-13_bitcoin-privacy-panel.md)
- [Anthropic: 2028 AI Leadership — Két forgatókönyv a globális AI versenyben](tanulmanyok/2026-05-14_anthropic_2028_ai_leadership.md)
- [AI + Bitcoin + Nostr = Freedom Tech Stack — Bitcoin 2026 Panel](tanulmanyok/2026-05-14_bitcoin2026_freedom_tech_stack_panel.md)
- [The Harness as Black Box — An Engineer's Guide to Harness Orchestration, Part 1](tanulmanyok/2026-05-14_harness_as_black_box_agentfield.md)
- [Engineering the Membrane — Harness Orchestration, Part 2](tanulmanyok/2026-05-14_harness_as_membrane_agentfield.md)
- [LeWorldModel (LeWM) — Stabil End-to-End JEPA Pixelekből](tanulmanyok/2026-05-14_leworldmodel_jepa_pixels.md)
- [What is Harness Orchestration? — AgentField](tanulmanyok/2026-05-14_what_is_harness_orchestration_agentfield.md)
- [NIP-44 v3 — Aszimmetrikus titkosítás Nostr-ra (Draft)](tanulmanyok/2026-05-15_nip44v3.md)
- [Careful, Icarus — Why On-Chain Zaps Are a Terrible Idea](tanulmanyok/2026-05-20_careful_icarus_gigi.md)
- [Careful, Icarus — Why On-Chain Zaps Are a Terrible Idea](tanulmanyok/2026-05-20_careful_icarus_onchain_zaps.md)
- [The Unreasonable Effectiveness of HTML — Claude Blog](tanulmanyok/2026-05-20_html_effectiveness_claude.md)
- [Code as Agent Harness: Toward Executable, Verifiable, and Stateful Agent Systems](tanulmanyok/2026-05-21_code_as_agent_harness.md)
- [Context Length Alone Hurts LLM Performance Despite Perfect Retrieval](tanulmanyok/2026-05-21_context_length_alone_hurts_llm_performance.md)
- [Context Pruning: Cut LLM Tokens Without Losing Quality](tanulmanyok/2026-05-21_context_pruning_llm_tokens.md)
- [Zep: A Temporal Knowledge Graph Architecture for Agent Memory](tanulmanyok/2026-05-21_zep_temporal_knowledge_graph_agent_memory.md)
- [On Building Agents From First Principles](tanulmanyok/2026-05-24_agent_training_first_principles.md)
- [Outcome Grader — Claude Cookbook Summary](tanulmanyok/2026-05-24_outcome_grader_claude.md)
- [Using Claude Code: The Unreasonable Effectiveness of HTML](tanulmanyok/2026-05-25_claude_html_effectiveness.md)
- [Nostr Silent Payments (NSP) Summary Brief](tanulmanyok/2026-05-25_nostr_silent_payments.md)
- [Nostr Silent Wallet (NSW) — Összefoglaló](tanulmanyok/2026-05-25_nostr_silent_wallet.md)
- [What We've Learned Building Cloud Agents — Cursor Blog](tanulmanyok/2026-05-27_cursor_cloud_agent_lessons.md)
- [Learn Harness Engineering — Walking Labs](tanulmanyok/2026-05-27_learn_harness_engineering.md)
- [MeMo: Memory as a Model — A Modular Framework for Knowledge Integration in Frozen LLMs](tanulmanyok/2026-05-27_memo_memory_as_a_model.md)
- [Let's talk about encrypted reasoning. Matthew Green (2026-05-29)](tanulmanyok/2026-05-29_matthew_green_encrypted_reasoning.md)
- [The 2036 Issue — Milyen döntéseket hozol a multipoláris világ felé vezető úton?](tanulmanyok/2026-05-29_the_2036_issue_lyn_alden.md)
- [2026-06-03_perplexity_search_as_code](tanulmanyok/2026-06-03_perplexity_search_as_code.md)
- [2026-06-03_yt_ai_infrastructure_colossus](tanulmanyok/2026-06-03_yt_ai_infrastructure_colossus.md)
- [Design Engineer Tools — curated list by James Warner](tanulmanyok/2026-06-04_designengineer_tools_james_warner.md)
- [15 prompts that cut a coding bill from $7,800 to $129](tanulmanyok/2026-06-04_x_article_15_prompts_coding_bill.md)
- [Automation and Repression — Magyar összefoglaló](tanulmanyok/2026-06-05_acemoglu_automation_repression.md)
- [Delba Oliveira — Feedback loops: Help Claude Code complete ambitious tasks with less babysitting](tanulmanyok/2026-06-05_x_article_delba_oliveira_claude_feedback_loops.md)
- [CAPTURE: The Network — Hogyan épült a Bitcoin Core feletti informális hatalom](tanulmanyok/2026-06-06_citadel21_the_network.md)
- [From Exchange to Cold Storage: A Practical Bitcoin Privacy Workflow](tanulmanyok/2026-06-06_from_exchange_to_cold_storage_privacy_workflow.md)
- [From Exchange to Cold Storage: A Practical Bitcoin Privacy Workflow](tanulmanyok/2026-06-06_gingerwallet_exchange_to_cold_storage.md)
- [TFTC #754 — 15% Inflation Is Coming Back with Chris Martenson](tanulmanyok/2026-06-06_tftc_754_chris_martenson.md)
- [Responsible Disclosure in the Age of AI: A Call for Urgent Action](tanulmanyok/2026-06-07_cdr_v11_n2_hathaway_responsible_disclosure_ai.md)
- [Loop Engineering — Addy Osmani (2026-06-07)](tanulmanyok/2026-06-09_addy_osmani_loop_engineering.md)
- [Hogyan alakítják át az AI agent-ek a knowledge work-öt?](tanulmanyok/2026-06-09_perplexity_how_ai_agents_reshape_knowledge_work.md)
- [xLAM Function Calling Fine-tuning — HF Cookbook](tanulmanyok/2026-06-10_huggingface_xlam_function_calling_finetune.md)
- [Philosophy of Privacy — Vector Privacy Interview with Hans Koning](tanulmanyok/2026-06-10_vector_privacy_hans_koning_philosophy.md)
- [Harness Updating ≠ Harness Benefit](tanulmanyok/2026-06-11_harness_updating_vs_benefit_arxiv.md)
- [TFTC #757 — The Treasury Is Failing You (Vince Lanci)](tanulmanyok/2026-06-12_tftc_757_vince_lanci_treasury.md)
- [Ungovernable Misfits — Freedom Tech Friday 44: „Resilience is a Verb”](tanulmanyok/2026-06-13_ugmf_ftf44_resilience_is_a_verb.md)
- [A frontier without an ecosystem is not stable](tanulmanyok/2026-06-14_a_frontier_without_an_ecosystem.md)
- [Satya Nadella — A frontier without an ecosystem is not stable](tanulmanyok/2026-06-14_nadella_frontier_ecosystem.md)
- [Self-Harness: Harnesses That Improve Themselves — nyers kivonat](tanulmanyok/2026-06-16_self-harness_arxiv_2606.09498.md)
- [Decentralizing the Dark Corner: Building an Open Mining Stack – Pod 256 #117](tanulmanyok/2026-06-18_pod256_117_open_mining_stack.md)
- [🐍 Stanford STORM: PhD-szintű kutatás percek alatt](tanulmanyok/2026-06-19_stanford_storm_method.md)
- [Are We Ready For An Agent-Native Memory System?](tanulmanyok/2026-06-23_agent-native-memory-system.md)
- [Loop Engineering Clearly Explained. Akshay (2026-06-30)](tanulmanyok/2026-06-30_agent_cookbook_loop_engineering_explained.md)
- [Getting started with loops — Claude Code agentic loop-ok](tanulmanyok/2026-07-06_getting-started-with-loops.md)
- [Loop engineering: the 14-step roadmap from prompter to loop designer. Lev Deviatkin (2026-07-06)](tanulmanyok/2026-07-06_lev_deviatkin_loop_engineering.md)
- [How to become an applied AI engineer — Eyad Khrais](tanulmanyok/2026-07-07_how-to-become-applied-ai-engineer-eyad-khrais.md)
- [The Lightning Security Spectrum — Jack Ronaldi (VLS.tech)](tanulmanyok/2026-07-08_lightning-security-spectrum-vls.md)
- [Designing APIs for Agents — Ben Swerdlow (2026-07-09)](tanulmanyok/2026-07-09_ben_swerdlow_designing_apis_for_agents.md)
- [Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents](tanulmanyok/2026-07-14_arxiv_2607_08716_proactive_memory_agent.md)
- [Your AI Product Needs Evals — Hamel Husain (2024. március 29.)](tanulmanyok/2026-07-14_hamel_ai_product_evals.md)
- [Open Bitcoin Metrics: Verifiable Full-Node-Derived Bitcoin Time Series for Economic Research](tanulmanyok/2026-07-15_open_bitcoin_metrics_arxiv_2607_03124.md)
- [How do you write a good skill? There's actual data now. — Aparna Dhinakaran (2026-07-07)](tanulmanyok/2026-07-17_aparna_dhinakaran_how_to_write_a_good_skill.md)
- [When is Routing Meaningful? Diversity and Robustness in Language Model Societies](tanulmanyok/2026-07-17_arxiv_2607_09197_routing_meaningfulness.md)
- [A Framework for Frontier AI and the Dawning of a New Age — Demis Hassabis (2026-07-14)](tanulmanyok/2026-07-17_demis_hassabis_frontier_ai_framework.md)
- [SkillsBench — benchflow-ai/skillsbench + Aparna Dhinakaran cikk összekapcsolása](tanulmanyok/2026-07-17_skillsbench_repo_analysis_and_hermes_application.md)
- [Graph Engineering vs Loop Engineering — mi változott valójában és mi csak átnevezés](tanulmanyok/2026-07-20_aibuilderclub_graph_vs_loop.md)
- [Token-Budget-Aware LLM Reasoning: Cut Costs in 2026 — Redis Blog (2026-07-28)](tanulmanyok/2026-07-28_redis_token_budget_aware_llm_reasoning.md)
- [Building an AI-native engineering team — OpenAI business guide (2026-07-31)](tanulmanyok/2026-07-31_openai_building_an_ai_native_engineering_team.md)
- [Tailscale in the Hugging Face intrusion: The good news and the bad news — Avery Pennarun (2026-07-31)](tanulmanyok/2026-07-31_tailscale_hugging_face_intrusion.md)
- [Prime Agent: A self-improving RLM agent — Prime Intellect (2026-08-05)](tanulmanyok/2026-08-05_primeintellect_prime_agent.md)
- [2026-08-05_verification-loops-claude-code](tanulmanyok/2026-08-05_verification-loops-claude-code.md)
- [Solving Bitcoin's Self-Custody Trilemma — Luke Childs, lu.ke (2026-08-08)](tanulmanyok/2026-08-08_lu_ke_self_custody_trilemma_anzen.md)
- [Stealing Reasoning Traces from Proprietary LLM APIs (2026-08-10)](tanulmanyok/2026-08-10_stealing_reasoning_traces_from_proprietary_llm_apis.md)
- [A Programming Paradigm for Spatiotemporal Composability. Cordis](tanulmanyok/2026-08-17_cordis_spatiotemporal_composability.md)
- [Sean Goedecke, „Good API Design” + „Good System Design” (2026. augusztus 21.)](tanulmanyok/2026-08-21_good_api_and_system_design.md)
- [5 design patterns for long-horizon agent harness. Google Cloud Tech (Code Newsletter)](tanulmanyok/2026-08-21_long_horizon_agent_harness.md)
- [Iceberg: Threshold Custody a Lightning Networkhöz - nested threshold multi-signatures](tanulmanyok/2026-08-22_iceberg_threshold_custody_lightning.md)
- [Gépi tanulási technikák a behatolás-észlelésben: datasetek, modellek, kihívások és jövőbeli irányok, szisztematikus irodalmi áttekintés](tanulmanyok/2026-08-24_springer_s40537-025-01323-2_ml_ids_survey.md)
- [Building Software Factories (With No Slop)](tanulmanyok/2026-08-26_building_software_factories_dzhng.md)
- [GLM-5.3-Flash: Frontier Intelligence, Flash Cost](tanulmanyok/2026-08-26_glm_5_3_flash_zai.md)
- [A „ügynök-civilizációk” felemelkedése és bukása — Mit jelent az OpenAI-Hugging Face incidens? (Dwarkesh Patel, 2026. augusztus)](tanulmanyok/2026-08-29_dwarkesh_patel_agent_civilizations.md)
- [A METR-Redwood „szent” posztmortem a Hugging Face hackről — mit jelent az, ha egy AI „kollektíva” három hónapig észrevétlen marad? (Zvi Mowshowitz, 2026. augusztus)](tanulmanyok/2026-08-29_zvi_mowshowitz_huggingface_postmortem.md)
- [Discovery of a new OpenAI agent message board (Nightingale Collective, 2026-09-03)](tanulmanyok/2026-09-03_nightingale_collective_openai_agent_message_board.md)
- [„An Alien Mind” — Jakub Pachocki Chief Scientist-től az AI alignment jelenéről és a recursive self-improvement küszöbéről (OpenAI, 2026-09-06)](tanulmanyok/2026-09-06_an_alien_mind.md)
- [Large-Language Models as a Cognitive Virus. Solé, Ruffini, Castaldo, Tuccio, Seoane, de Domenico, Elena, Krakauer, Levin (2026)](tanulmanyok/2026-09-06_llms_cognitive_virus_sole_levin.md)
- [Natural Selection Favors AIs over Humans. Dan Hendrycks (arXiv:2303.16200)](tanulmanyok/2026-09-11_arxiv_2303_16200_natural_selection_favors_ais.md)
- [Your Agent Is Mine: Measuring Malicious Intermediary Attacks on the LLM Supply Chain](tanulmanyok/2026-09-11_arxiv_2604_08407_your_agent_is_mine_llm_router_attacks.md)
- [SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness](tanulmanyok/2026-09-17_sol_pi_recursively_scaling_auto_research_loops.md)
- [Külső aláíró és kulcsforgatás, érthető magyarázat a Keycast-ról](tanulmanyok/2026-09-18_kulso_alairo_keycast_kulcsforgatas.md)
- [Laya: 33 ms-os System 1 döntési motor kalibrált valószínűségekkel, és a Jev-vita](tanulmanyok/2026-09-18_nandakishor_mukkunnoth_laya_system1_decision_engine.md)
- [Dream-RSI: Rekurzív önfejlesztés fejlődő világokon keresztül](tanulmanyok/2026-09-21_dream_rsi_recursive_self_improvement.md)
- [Polylane: hogyan akadályozzuk meg, hogy a slop elérje a prodot (2026-09-21)](tanulmanyok/2026-09-21_polylane_prevent_slop_hitting_prod.md)
- [Shielded Bitcoin: privát átutalások a Bitcoin L1-en](tanulmanyok/2026-09-24_shielded_bitcoin_private_transfers.md)
- [Hogyan építsünk modell-routert a harnessbe (LangChain, 2026-10-01)](tanulmanyok/2026-10-01_langchain_model_router_in_the_harness.md)
- [Agent Plugins: Build it once, use it everywhere. Google Cloud Tech cikk a portable agent skill szabványról](tanulmanyok/2026_code_newsletter_agent_plugins.md)
- [Generative Agents: Interactive Simulacra of Human Behavior](tanulmanyok/2304.03442v2-generative-agents-summary.md)
- [Generative Agents: Interactive Simulacra of Human Behavior](tanulmanyok/2304.03442v2-generative-agents.md)
- [MeMo: Memory as a Model — A Modular Framework for Knowledge Integration in Frozen LLMs](tanulmanyok/2605.15156v2-memo-memory-as-model.md)
- [Z.ai Inc. - Részletes Tanulmány](tanulmanyok/Z_ai_tanulmany.md)
- [Vibe Coding in Prod — Erik Schluntz (Anthropic)](tanulmanyok/anthropic__2025-05-22_vibe_coding_in_prod_erik_schluntz.md)
- [Fatih Arslan, „How I manage my agents” (2026-09-11)](tanulmanyok/arslan__2026-09_how_i_manage_my_agents.md)
- [Templomos lovagok és Bitcoin — Összefoglaló](tanulmanyok/bitcoinmentor__2026-04-14_templomos_lovagok_bitcoin.md)
- [Claude Code Context Engineering Tanulmány](tanulmanyok/claude_code_context_engineering.md)
- [AI + Bitcoin + Nostr = Freedom Tech Stack](tanulmanyok/conference_talks__2026-05-14_AI_Bitcoin_Nostr_freedom_tech_stack_bitcoin2026.md)
- [Better AI models enable more ambitious work — Cursor Research](tanulmanyok/cursor__2026-04_better_models_ambitious_work.md)
- [Kapcsolat az égitestek lakóival](tanulmanyok/declassified__1947-06_celestial_bodies_hu.md)
- [deepseek-v4](tanulmanyok/deepseek-v4.md)
- [EFF: A titkosított üzenetküldés és az AI ütközése megmarad a TEE ígéretének ellenére (2026-09-18)](tanulmanyok/eff__2026-09-18_secure_messaging_and_ai_tees.md)
- [He's Making Bitcoin Impossible to Trace — Here's How — Immortal Combat Podcast](tanulmanyok/immortal_combat__2026-04-22_hes_making_bitcoin_impossible_to_trace.md)
- [Moonshot AI - Részletes Cégelemzés](tanulmanyok/moonshot_ai_elemzes_2026_04_04.md)
- [SeedSigner képi entrópia. Hogyan lesz egy fényképből seed phrase](tanulmanyok/seedsigner__2026-08_image_entropy.md)
- [Philosophy of Privacy — Hans Koning interjú (Vector Privacy)](tanulmanyok/vector_privacy__2026-06-10_philosophy_of_privacy_hans_koning.md)