cs.AIMar 17, 2026

Interpretable Context Methodology: Folder Structure as Agentic Architecture

Authors: Jake Van Clief, David McDermott

Organizations: Eduba, University of Edinburgh, Palm Coast, Florida, USA

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Current approaches to AI agent orchestration typically involve building multi-agent frameworks that manage context passing, memory, error handling, and step coordination through code. These frameworks work well for complex, concurrent systems. But for sequential workflows where a human reviews output at each step, they introduce engineering overhead that the problem does not require. This paper presents Model Workspace Protocol (MWP), a method that replaces framework-level orchestration with filesystem structure. Numbered folders represent stages. Plain markdown files carry the prompts and context that tell a single AI agent what role to play at each step. Local scripts handle the mechanical work that does not need AI at all. The result is a system where one agent, reading the right files at the right moment, does the work that would otherwise require a multi-agent framework. This approach applies ideas from Unix pipeline design, modular decomposition, multi-pass compilation, and literate programming to the specific problem of structuring context for AI agents. The protocol is open source under the MIT license.

Figures & tables

DimensionFramework approachICM approach
Change stage orderEdit orchestration code, redeployRename or reorder folders
Modify a promptEdit agent configuration in codeEdit a markdown file
Add or remove a stageWrite new agent class, update orchestratorAdd or delete a folder
Inspect intermediate stateAdd logging, build dashboardOpen the folder, read the files
Hand off to another personDocument environment, dependencies, setupCopy the folder
Who can make changesDeveloperAnyone with a text editor
Table 1. Comparison of control surfaces for sequential, human-reviewed workflows. The first six rows show dimensions where ICM’s filesystem approach simplifies common operations. The last four rows show dimensions where framework-based approaches provide capabilities that ICM lacks or handles less well.
Figure 1. The five-layer context hierarchy. Layers 0–2 provide structural routing and stage instructions. Layers 3 and 4 carry content: Layer 3 holds reference material (the factory), stable across runs; Layer 4 holds working artifacts (the product), unique to each run.
 Layer 3: ReferenceLayer 4: Working
Changes between runsNoYes
Example filesvoice.md, design-system.md, conventions.mdresearch-output.md, script-draft.md
Model shouldInternalize as constraintsProcess as input
Configured duringWorkspace setup (once)Pipeline execution (each run)
Folder locationreferences/ , _config/ , shared/output/
AnalogyThe recipeThe ingredients
Table 2. Layer 3 (reference material) versus Layer 4 (working artifacts).
Figure 2. Folder structure of a typical ICM workspace, with layer annotations. Files and folders are color-coded by their role in the context hierarchy. Layer 3 material (reference) persists across runs. Layer 4 material (working artifacts) changes each time the pipeline executes.
Figure 3. Context window composition by stage (representative token counts from the script-to-animation workspace). The three ICM stages each deliver 2,000–8,000 focused tokens. A monolithic approach loading all stages’ instructions, all reference material, and all prior outputs produces a context window exceeding 40,000 tokens, most of it irrelevant to the current task.
Figure 4. Pipeline flow through three stages with review gates. Each stage receives its own context (Layers 0–4), writes output to its folder, and the human reviews and optionally edits before the next stage reads it. The same model executes every stage; the folder structure controls what context it receives.
Figure 5. Observed frequency of human edits at each stage boundary, reported by 33 practitioners using multi-stage ICM workspaces. Intervention follows a U-shaped pattern: heavy at stage 1 (direction-setting), light at middle stages (constrained execution), heavy again at the final stage (aligning output with earlier decisions). Stage 1 editing is creative judgment. Final-stage editing is closer to debugging. Values are approximate and based on practitioner self-report through conversation, not instrumented measurement.
More in this topic

LLM Agent Orchestration

Large language model agent orchestration coordinates models, tools, and workflows to complete complex tasks reliably and efficiently.

See all 458 papers

Explore similar work

CardsList

  1. Separating Intelligence from Execution: A Workflow Engine for the Model Context Protocol

    Mar 13, 2026Abhinav Singh ParmarLLM Agent Tool GenerationLLM Agent Orchestration

    A Preprint

  2. AWCP: A Workspace Delegation Protocol for Deep-Engagement Collaboration across Remote Agents

    Feb 24, 2026Xiaohang Nie, Zihan Guo, Youliang Chen +2Multi-Agent CollaborationMulti-Agent LLM Systems

    Harbin Institute of Technology · Shanghai Innovation Institute · Sun Yat-sen University +2

  3. WorkflowOps: Learning Agent Collaboration Priors for Multi-Agent Workflow Orchestration

    Oct 6, 2026Qi Cheng, Shengyu Chen, Wei Cheng +4LLM Agent OrchestrationLLM Agent Workflow Optimization

    Rutgers University, New Brunswick, NJ · University of Pittsburgh, Pittsburgh, PA · NEC Labs America, Princeton, NJ