Control-Plane Placement and Memory Substrates in Modern Agentic AI: A Comparative Method for Prompt-Native Systems, Agent Shells, and Durable Runtimes

A. Bondarenko; N. Zamorskaia; D. Djakons; A. Djakona; R. Djakons

doi:10.2478/lpts-2026-0036

.blurhash-client-img { display: none !important; }

Control-Plane Placement and Memory Substrates in Modern Agentic AI: A Comparative Method for Prompt-Native Systems, Agent Shells, and Durable Runtimes

Latvian Journal of Physics and Technical Sciences

Volume 63 (2026): Issue 5 (October 2026)

By: A. Bondarenko, N. Zamorskaia, D. Djakons, A. Djakona and R. Djakons

Open Access

|Sep 2026

Abstract

AI agents that autonomously write, test, and modify code are now deployed across architectures that share similar surface features but differ substantially in where and how operating control is exercised. Contemporary systems span general-purpose SDKs, prompt-native work-flow packs, interactive coding shells, hosted agents, memory services, and durable workflow runtimes, yet the existing language of framework comparison often treats them as interchangeable. The remaining problem is comparative precision: these systems still look artificially similar if operational control and durable experience are assumed to live in the same architectural layer. This paper proposes control-plane placement, complemented by memory substrate analysis, as a comparative method for modern agentic systems. The organising principle is the enforcement set – the collection of architectural layers that can independently constrain agent behaviour, and the five system families compared here represent its five stable configurations. The method asks where workflow policy, tool exposure, permissioning, routing, verification, and recovery are specified and enforced, and separately how durable experience is stored, searched, updated, and reused across sessions. Using official documentation and open-source implementation evidence, the paper compares prompt-native systems, prompt-modular shells, runtime-governed shells, host-mediated or coordinator forms, and durable workflow runtimes across a twelve-system ecosystem. It contributes a qualitative coding matrix for control-plane salience and a companion memory-oriented comparison covering artefact memory, procedural memory, write-policy visibility, and traceability. The paper also argues that coding-agent benchmarks are best read as evaluations of model × reasoning policy × harness/runtime × memory substrate × environment under explicit tool, budget, and infrastructure conditions. The result is a more discriminating vocabulary for agentic AI and a clearer method for comparing rapidly converging systems.

References

Liu, J. (2026, March 3). LlamaIndex is more than a RAG framework: It is agentic document processing. LlamaIndex. https://www.llamaindex.ai/blog/llamaindex-is-more-than-a-rag-framework.
Search in Google Scholar Back to article
Pan, L., Zou, L., Guo, S., Ni, J., & Zheng, H.-T. (2026). Natural-language agent harnesses [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2603.25723
Search in Google Scholar Back to article
Zhou, C., Chai, H., Chen, W., Guo, Z., Shan, R., Song, Y., . . . & Zhang, W. (2026). Externalization in LLM agents: A unified review of memory, skills, protocols and harness engineering [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2604.08224
Search in Google Scholar Back to article
Yue, L., Bhandari, K. R., Ko, C.-Y., Patel, D., Lin, S., Zhou, N., … & Pan, S. (2026). From static templates to dynamic runtime graphs: A survey of workflow optimization for LLM agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2603.22386
Search in Google Scholar Back to article
Rombaut, B. (2026). Inside the scaffold: A source-code taxonomy of coding agent architectures. [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2604.03515
Search in Google Scholar Back to article
Anthropic. (2026, January 9). Demystifying evals for AI agents. https://www.anthropic.com/engineering/demystifying-evalsfor-ai-agents.
Search in Google Scholar Back to article
Anthropic. (2025–2026). Claude Code materials, including How Claude Code works, Create custom subagents, Agent SDK overview, and Modifying system prompts. https://code.claude.com/docs/en/how-claude-code-works (documentation set).
Search in Google Scholar Back to article
GitHub / Microsoft / VS Code team. (2025–2026). Copilot agent-mode and coding-agent materials, including Introducing GitHub Copilot agent mode (preview), Local agents in Visual Studio Code, and About GitHub Copilot cloud agent. https://code.visualstudio.com/blogs/2025/02/24/introducing-copilot-agent-mode (documentation set).
Search in Google Scholar Back to article
OpenAI. (2025–2026). Agents SDK and workflow tooling materials, including the Building agents track, Agents SDK guide, Agent Builder guide, and Results and state guide. http://developers.openai.com/tracks/building-agents (documentation set).
Search in Google Scholar Back to article
Google. (2025–2026). Agent Development Kit (ADK) materials, including workflow agents, callbacks, and skills documentation. https://google.github.io/adk-docs/agents/workflow-agents/ (documentation set).
Search in Google Scholar Back to article
Cline. (2025–2026). Documentation and repository materials, including Plan & Act docs and prompt architecture. https://docs.cline.bot/ (see also the project repository at https://github.com/cline/cline/)
Search in Google Scholar Back to article
Anomalyco. (2026). OpenCode repository and session-runtime sources. https://github.com/anomalyco/opencode
Search in Google Scholar Back to article
QwenLM. (2025–2026). Qwen Code materials, including Approval Mode, Subagents, Settings, and repository docs. https://qwenlm.github.io/qwen-code-docs/en/users/features/approval-mode/ (documentation set).
Search in Google Scholar Back to article
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E. H., Le, Q. V., & Zhou, D. (2022). Chain-of-thought prompting elicits reasoning in large language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2201.11903
Search in Google Scholar Back to article
Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. (2022). ReAct: Synergizing reasoning and acting in language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2210.03629
Search in Google Scholar Back to article
Yao, S., Yu, D., Zhao, J., Shafran, I., Griffiths, T. L., Cao, Y., & Narasimhan, K. (2023). Tree of thoughts: Deliberate problem solving with large language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2305.10601
Search in Google Scholar Back to article
Besta, M., Blach, N., Kubicek, A., Gerstenberger, R., Podstawski, M., Gianinazzi, L., … & Hoefler, T. (2023). Graph of thoughts: Solving elaborate problems with large language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2308.09687
Search in Google Scholar Back to article
Madaan, A., Tandon, N., Gupta, P., Hallinan, S., Gao, L., Wiegreffe, S., … & Clark, P. (2023). Self-refine: Iterative refinement with self-feedback [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2303.17651
Search in Google Scholar Back to article
Shinn, N., Cassano, F., Berman, E., Gopinath, A., Narasimhan, K., & Yao, S. (2023). Reflexion: Language agents with verbal reinforcement learning [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2303.11366
Search in Google Scholar Back to article
Dhuliawala, S., Komeili, M., Xu, J., Raileanu, R., Li, X., Celikyilmaz, A., & Weston, J. (2023). Chain-of-verification reduces hallucination in large language models [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2309.11495
Search in Google Scholar Back to article
Fricke, F., Malberg, S., & Groh, G. (2026). Framework of thoughts: A foundation framework for dynamic and optimized reasoning based on chains, trees, and graphs. In Proceedings of the First Workshop on Structured Understanding, Retrieval, and Generation in the LLM Era (SURGeLLM 2026) (pp. 132–151). Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.surgellm-1.8
Search in Google Scholar Back to article
Plaat, A., Wong, A., Verberne, S., Broekens, J., van Stein, N., & Bäck, T. (2025). Multi-step reasoning with large language models: A survey. ACM Computing Surveys, 58(6), 1–35. https://doi.org/10.1145/3774896
Search in Google Scholar Back to article
Bondarenko, A., Zamorskaia, N., Djakons, D., Djakona, A., & Djakons, R. (2026). Rapid evolution of agentic AI: From framework labels to control surfaces [Short overview]. Riga Nordic University.
Search in Google Scholar Back to article
SWE-bench project. (2024–2026). Official benchmark materials, including the SWE-bench leaderboard and mini-SWE-agent. https://www.swebench.com/ (benchmark site and mini-SWE-agent materials).
Search in Google Scholar Back to article
Merrill, M. A., Shaw, A. G., Carlini, N., Li, B., Raj, H., Bercovich, I., . . . & Schmidt, L. (2026). Terminal-Bench: Benchmarking agents on hard, realistic tasks in command line interfaces [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.11868
Search in Google Scholar Back to article
Dao, M.-D., Le, Q. M., Lam, H. T., Le, D.-T., Pham, Q.-V., O’Sullivan, B., & Nguyen, H. D. (2026). Agentic design patterns: A system-theoretic framework [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.19752
Search in Google Scholar Back to article
Milosevic, Z., & Rabhi, F. (2026). Architecting agentic communities using design patterns [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.03624
Search in Google Scholar Back to article
Xu, W., Wang, Y., Mei, K., Liang, K., Wang, Z., Jin, M., … & Metaxas, D. N. (2026). MemGym: A long-horizon memory environment for LLM agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2605.20833
Search in Google Scholar Back to article
Wu, D., Wang, H., Yu, W., Zhang, Y., Chang, K.-W., & Yu, D. (2024). LongMemEval: Benchmarking chat assistants on long-term interactive memory [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2410.10813
Search in Google Scholar Back to article
Maharana, A., Lee, D.-H., Tulyakov, S., Bansal, M., Barbieri, F., & Fang, Y. (2024). Evaluating very long-term conversational memory of LLM agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2402.17753
Search in Google Scholar Back to article
He, Z., Wang, Y., Zhi, C., Hu, Y., Chen, T.-P., Yin, L., … & Pentland, A. (2026). MemoryArena: Benchmarking agentmemory in interdependent multi-session agentic tasks [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2602.16313
Search in Google Scholar Back to article
Shen, Y., Li, K., Zhou, W., & Hu, S. (2026). Mem2ActBench: A benchmark for evaluating long-term memory utilization in task-oriented autonomous agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.19935
Search in Google Scholar Back to article
Li, X., Liu, Y., Chen, W., You, B., Di, Z., He, Y., . . . & Song, D. (2026). SkillsBench: Benchmarking how well agent skills work across diverse tasks [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2602.12670
Search in Google Scholar Back to article
Zhu, J., Wu, J., Hu, M., Zhu, S., Pan, J., Shen, W., … & Xu, M. (2026). SWE Context Bench: A benchmark for context learning in coding [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2602.08316
Search in Google Scholar Back to article
pi.dev/Zechner, M. (2025–2026). Pi coding agent materials, including the official site and architectural write-up on explicit artifact memory. https://pi.dev/ (see also https://mariozechner.at/posts/2025-11-30-pi-coding-agent/)
Search in Google Scholar Back to article
Nous Research. (2026). Hermes Agent official materials. https://hermes-agent.nousresearch.com/
Search in Google Scholar Back to article
Honcho. (2026). Honcho design patterns and memory-service architecture materials. https://honcho.dev/docs/v3/documentation/core-concepts/design-patterns
Search in Google Scholar Back to article
Sumers, T. R., Yao, S., Narasimhan, K., & Griffiths, T. L. (2023). Cognitive architectures for language agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2309.02427
Search in Google Scholar Back to article
Packer, C., Wooders, S., Lin, K., Fang, V., Patil, S. G., Stoica, I., & Gonzalez, J. E. (2023). MemGPT: Towards LLMs as operating systems [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2310.08560
Search in Google Scholar Back to article
Du, P. (2026). Memory for Autonomous LLM Agents: Mechanisms, evaluation, and emerging frontiers. [Preprint]. arXiv. https://arxiv.org/abs/2603.07670
Search in Google Scholar Back to article
Nous Research. (2026). Hermes Agent memory documentation. https://hermes-agent.nousresearch.com/docs/user-guide/features/memory
Search in Google Scholar Back to article
Fang, J., Xu, B., Wang, Z., Cao, H., Deng, X., Dong, B., … & Zhang, N. (2026). Rethinking memory as continuously evolving connectivity [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2605.28773
Search in Google Scholar Back to article
Chen, H., Zhang, Z., Pei, W., He, B., Wu, M., Zeng, J., … & Zhang, H. (2026). MemForest: An efficient agent memory system with hierarchical temporal indexing [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2605.23986
Search in Google Scholar Back to article
Latimer, C., Boschi, N., Neeser, A., Bartholomew, C., Srivastava, G., Wang, X., & Ramakrishnan, N. (2025). Hindsight is 20/20: Building agent memory that retains, recalls, and reflects [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2512.12818
Search in Google Scholar Back to article
Yu, Y., Yao, L., Xie, Y., Tan, Q., Feng, J., Li, Y., & Wu, L. (2026). Agentic memory: Learning unified long-term and short-term memory management for large language model agents [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.01885
Search in Google Scholar Back to article
OpenClaw project. (2026). OpenClaw repository and platform materials. https://github.com/openclaw/openclaw
Search in Google Scholar Back to article
obra. (2025–2026). superpowers repository. https://github.com/obra/superpowers
Search in Google Scholar Back to article
glebis. (2025–2026). claude-skills repository. https://github.com/glebis/claude-skills
Search in Google Scholar Back to article
Anthropic. (2025–2026). skills repository and marketplace materials. https://github.com/anthropics/skills
Search in Google Scholar Back to article
GitHub. (2025–2026). awesome-copilot repository. https://github.com/github/awesome-copilot
Search in Google Scholar Back to article
LangChain. (2025–2026). LangGraph materials, including the LangGraph overview, persistence docs, and workflowsand-agents documentation. https://docs.langchain.com/oss/python/langgraph/overview (documentation set).
Search in Google Scholar Back to article
Deus Data. (2026). codebase-memory-mcp: High-performance code intelligence MCP server with persistent knowledge graph. https://github.com/DeusData/codebase-memory-mcp (see also the preprint at https://arxiv.org/abs/2603.27277).
Search in Google Scholar Back to article
Shamsi, S. (2026). graphify: AI coding assistant skill for building queryable knowledge graphs from any codebase. [Computer software]. GitHub. https://github.com/safishamsi/graphify.
Search in Google Scholar Back to article
CocoIndex. (2026). cocoindex-code: Lightweight embedded AST-based code search engine and MCP server for coding agents. [Computer software]. GitHub. https://github.com/cocoindex-io/cocoindex-code.
Search in Google Scholar Back to article
Li, X., Ben-Israel, N., Raz, Y., Ahmed, B., Serebro, D., & Raux, A. (2026). RepoMod-Bench: A benchmark for code repository modernization via implementation-agnostic testing [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2602.22518
Search in Google Scholar Back to article
Heule, S., Jia, E., & Jain, N. (2025, November 6). Improving agent with semantic search. Cursor. https://cursor.com/blog/semsearch
Search in Google Scholar Back to article
Agarwal, A., & Peet, C. (2025, November 19). How we’re making GitHub Copilot smarter with fewer tools. The GitHub Blog. https://github.blog/ai-and-ml/githubcopilot/how-were-making-github-copilot-smarter-with-fewer-tools/
Search in Google Scholar Back to article
Anthropic. (2026, February 5). Quantifying infrastructure noise in agentic coding evals. https://www.anthropic.com/engineering/infrastructure-noise
Search in Google Scholar Back to article
Wu, D., Ji, Z., Kawatkar, A., Kwan, B., Gu, J.-C., Peng, N., & Chang, K.-W. (2026). LongMemEval-V2: Evaluating long-term agent memory toward experienced colleagues [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2605.12493
Search in Google Scholar Back to article
Wang, H., Mang, Q., Cheung, A., Sen, K., & Song, D. (2026, April 8). How we broke top AI agent benchmarks: And what comes next. Berkeley RDI. https://rdi.berkeley.edu/blog/trustworthy-benchmarks-cont/
Search in Google Scholar Back to article
Anthropic. (2024, November 25). Introducing the Model Context Protocol. https://www.anthropic.com/news/model-context-protocol
Search in Google Scholar Back to article
Zietsman, C. (2026). The specification as quality gate: Three hypotheses on AI-assisted code review [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2603.25773
Search in Google Scholar Back to article
Żywot, A., Chen, X., Yuan, Y., Søgaard, A., & de Rijke, M. (2026). Can small agents collaborate to beat a single large language model? [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2601.11327
Search in Google Scholar Back to article
Kulkarni, S., & Kulkarni, Y. (2026). Benchmarking multi-agent LLM architectures for financial document processing: A comparative study of orchestration patterns, cost-accuracy tradeoffs and production scaling strategies [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2603.22651
Search in Google Scholar Back to article

Articles in this issue

DOI: https://doi.org/10.2478/lpts-2026-0036 | Journal eISSN: 2255-8896 (formerly 0868-8257) | Journal ISSN: 0868-8257

Journal RSS Feed

Language: English

Page range: 46 - 64

Published online: Sep 21, 2026

Published in print: Oct 1, 2026

Published by: Institute of Physical Energetics

In partnership with: Paradigm Publishing Services

Keywords:

control-plane placement,

Related subjects:

Physics,

Technical and applied physics

© 2026 A. Bondarenko, N. Zamorskaia, D. Djakons, A. Djakona, R. Djakons, published by Institute of Physical Energetics
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Volume 63 (2026): Issue 5 (October 2026)