Large language models can generate fluent clinical case vignettes, but fluency alone does not ensure fidelity to a specifiable clinical structure. We introduce FORMA, a theory-grounded framework that compiles a cognitive model of a disorder into a directed weighted graph, samples a person-specific configuration of that graph, and validates whether the generated vignette preserves the specified components and causal links. We instantiate FORMA on Posttraumatic Stress Disorder using the Ehlers and Clark cognitive model, generating 16,500 vignettes across 500 personas, 11 generation models, and three ablation conditions. Evaluation combines an external edge-recovery probe, two clinical experts, a scaled LLM judge, and a clinician user study with 100 licensed practitioners. The cognitive graph is recoverable from full-condition vignettes (MCC = +0.41, AUC = 0.70) but not from zero-shot generation (MCC = +0.01, AUC = 0.50). Experts rate full vignettes substantially higher than zero-shot alternatives, and clinicians perceive them to be human-written 85% of the time, compared with 22% for zero-shot. FORMA also reduces demographic disparity in perceived quality by 1.5–7×. These results show that cognitive formulation can serve as an auditable specification for scalable synthetic clinical text generation. A repository with the data and code is available online: https://github.com/Amit-Oren/FORMA.

Large language models are increasingly deployed in multi-turn settings such as tutoring, support, and counseling, where reliability depends on preserving consistent roles, personas, and goals across long horizons. This requirement becomes critical when LLMs are used to generate synthetic dialogues for training and evaluation, since LLM–LLM conversations can accumulate identity-related failures such as persona drift, role confusion, and “echoing”, where one agent gradually mirrors its partner. We introduce SPASM (Stable Persona-driven Agent Simulation for Multi-turn dialogue generation), a modular, stability-first framework that decomposes simulation into (i) persona creation via schema sampling, plausibility validation, and natural-language persona crafting, (ii) Client–Responder dialogue generation, and (iii) termination detection for coherent stopping. To improve long-horizon stability without changing model weights, we propose Egocentric Context Projection (ECP): dialogue history is stored in a perspective-agnostic representation and deterministically projected into each agent’s egocentric view before generation. Across three LLM backbones (GPT-4o-mini, DeepSeek-V3.2, Qwen-Plus) and nine Client–Responder pairings, we construct a dataset of 4,500 personas and 45,000 conversations (500 personas × 10 conversations per pairing). Ablations show ECP substantially reduces persona drift and, under human validation, eliminates echoing; embedding analyses recover persona structure and reveal strong responder-driven interaction geometry. Our code is available at https://github.com/lhannnn/SPASM.

LLM-based agents are increasingly deployed for mental-health support and crisis counselling, yet recent evaluations reveal that commercial therapy chatbots respond appropriately only about half the time in clinical scenarios.Clinicians and safety engineers are called upon to audit these systems, but existing tools do not surface the subtler psychosocial harms (manipulation, discrimination, psychological distress) nor produce the explainable rationales that practitioners need.We present DialogGuard, an open-source system that lets practitioners inspect, stress-test, and create audit trails for prompted LLM agents across five psychosocial safety dimensions.The system wraps around arbitrary generative models through four LLM-as-a-judge pipelines (single-agent scoring, dual-agent correction, multi-agent debate, and majority voting), each grounded in shared three-level rubrics.Through its web interface, practitioners evaluate agents in two modes (Live Chat and Manual Input) and review per-dimension risk scores with natural-language rationales.Experiments on PKU-SafeRLHF show that dual-agent correction provides the best accuracy-robustness trade-off, and a formative study with 12 practitioners confirms that the system supports prompt auditing, safety inspection, and supervisory decision-making.Code and demo: https://github.com/lhannnn/dialogguard-web.