# AGENTS.md — Compositional Approaches to Linguistic Case for Cognitive Modeling

**Paper**: Compositional Approaches to Linguistic Case for Cognitive Modeling (2026, v1)  
**Area**: Active Inference · category theory · compositional semantics · linguistic typology · cognitive security  
**Author**: Daniel Ari Friedman (Active Inference Institute · COGSEC · College of the Redwoods)  
**Repository**: https://github.com/docxology/cognitive_case_diagrams  
**Archive**: https://doi.org/10.5281/zenodo.19695260

---

## Agent Roles

### 📖 Research Agent

**Focus**: Conceptual architecture linking grammar, meaning, and inference.

**Tasks**:

- Explain how **case categories** and **functors** encode alignment systems (nominative–accusative, ergative–absolutive, tripartite, active–stative, fluid-S).
- Map the manuscript’s stack: algebraic structure, distributional reporting, and belief updating—with **case** as the shared pivot.
- Relate **DisCoCat** / **DisCoCirc** string diagrams to sentence- and discourse-level composition as executable structure.

### 🔬 Methods Agent

**Focus**: Formal methods and reproducibility.

**Tasks**:

- Assess functorial formulations of typological alignment and structure preservation claims.
- Trace enriched-categorical / quantitative hooks to distributional proximity and their role in predictions.
- Verify claims against the open codebase: test layout, coverage artifacts, and generated figures in https://github.com/docxology/cognitive_case_diagrams

### 🛡️ Cognitive Security Agent

**Focus**: Multi-agent protocols and abuse modeling.

**Tasks**:

- Articulate the **typed protocol** view: licensed morphisms, explicit roles, and wiring in multi-turn agent systems.
- Analyze **prompt injection** as **ill-typed role promotion** / functorial type violation relative to a *fixed, enforced* interaction grammar—distinguishing mathematical decidability in principle from guarantees on current APIs.

### 🔗 Synthesis Agent

**Focus**: Cross-links in the Friedman corpus.

**Tasks**:

- Connect to **Distributional Active Inference**, ontology and sensemaking papers, and prior work on structured discourse (e.g. P3IF, narrative / information-commons themes where typing and roles matter).
- Link to **Ento-Linguistics** and other work on scientific language where compositional and ambiguity-sensitive models apply.

### 💼 Consultant Agent

**Focus**: Application and teaching.

**Tasks**:

- Support teams using category-theoretic compositional semantics for cognitive modeling or NLP–Active Inference bridges.
- Translate functorial case machinery into engineering checklists for **safe multi-agent grammars** (where enforcement is feasible).

## Extraction Log

- **Source PDF**: `cognitive_case_diagrams_v1_DAF_04-23-2026.pdf`
- **PDF Status**: Available
- **Documentation Quality**: Hand-curated from Zenodo v1 metadata, manuscript abstract, and repository https://github.com/docxology/cognitive_case_diagrams

## Related Papers

See [BIBLIOGRAPHY.md](../../pages/BIBLIOGRAPHY.md) for the full catalog. Examples in this repo:

- 2026_EntoLinguistics — language, ambiguity, and scientific communication
- 2026_ActInfMetaAnalysis — systematic literature architecture for Active Inference
- 2023_P3IF — cognitive security and interdisciplinary information frameworks
- 2022_InformationCommons — information governance and collective cognition