mlsanityData to 14 Oct 2026

Dalek: A Constructive Agent Machine

arXiv 2609.03546 ↗First posted 3 Sep 2026cs.AIcited by 0

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

We present Dalek, a closed machine designed for agents that realizes self-maintenance, self-evolution, self-reproduction, and self-organization on any substrate satisfying a general host contract. The machine is built from three primitives---actors, messages, and channels. Four obligations---a host boundary, a construction language, admissible transitions, and rule heredity---give its boundary, identity, and closure a structural basis. Von Neumann's 1948 self-reproducing automaton supplies a hereditary constructional core: a self-description together with a constructor, a copier, and a controller. Dalek combines this core with the four obligations and rederives its medium for a text-and-message agent substrate, adding explicit structures for boundary, identity, history, and growth. A large language model and a compiler occupy the payload position and form a general capability producer. New capabilities are authored, compiled, installed into the description, and inherited by descendants. The same path produces the machine's own organs and even its runtime, closing heredity and evolution within the machine.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 14 of its 32 references that are in the map. First posted on arXiv 3 Sep 2026.

Builds on

14 of 32 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Metan: Recursive Self-Improvement through Emergent Depth25 Aug
  2. 2026Agent libOS: A Runtime Substrate for Capability-Controlled Self-Evolving LLM Agents2 Jun
  3. 2025Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents29 May
  4. 2025Building LLM Agents by Incorporating Insights from Computer Systems6 Apr

Other references · 10

  1. 2026Meta-Harness: End-to-End Optimization of Model Harnessescited by 309
  2. 2023Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generationcited by 148
  3. 2026Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnessescited by 142
  4. 2026Rethinking the Evaluation of Harness Evolution for Agentscited by 48
  5. 2023L2MAC: Large Language Model Automatic Computer for Extensive Code Generationcited by 30
  6. 2026Shepherd: Enabling Programmable Meta-Agents via Reversible Agentic Execution Tracescited by 12
  7. 2013Bounded Recursive Self-Improvementcited by 10
  8. 2026A Programming Paradigm for Spatiotemporal Composabilitycited by 8
  9. 2026Model-Native Computing Architecture: Envisioning Future System Architecture Through the Lens of Computer Architecturecited by 2
  10. 2026Self-Reference in Large Language Models: The Introspection Threshold for Recursive Self-Improvementcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026MOSS: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems
  2. 2026Critique of Agent Model
  3. 2026A Society of Researchers: Designing Institutions for Populations of Autonomous Research Agents
  4. 2026Towards the Harness of Embodied Agents
  5. 2026The Agent Use of Agent Beings: Agent Cybernetics Is the Missing Science of Foundation Agents
  6. 2026Drop the Hierarchy and Roles: How Self-Organizing LLM Agents Outperform Designed Structures
  7. 2024JaxLife: An Open-Ended Agentic Simulator
  8. 2026From Everything-is-a-File to Files-Are-All-You-Need: How Unix Philosophy Informs the Design of Agentic AI Systems