Skip to main content

​
Timepoint Pro

The first practical SNAG (Social Network Augmented Generation) engine. Simulates social dynamics by synthesizing and maintaining a structured social graph with causal provenance, knowledge flow, emotional states, and temporal consistency.

GitHub

timepoint-ai/timepoint-pro — Apache-2.0, Python, FastAPI + LangGraph

Detailed Docs

Full API reference, mechanisms, templates, and Python SDK docs

​
What SNAG Does Differently

RAGSNAG
Retrieves fromDocument chunksSocial graph state
TracksText similarityCausal chains, emotions, knowledge flow
TemporalSnapshot5 reasoning modes
EntitiesMentionsSimulated agents with state
OutputAnswer + citationsSimulation trajectory + dialog + provenance

​
5 Temporal Modes

ModeDescription
FORWARDStrict forward causality. Standard timelines.
PORTALBackward from target outcome. Goal decomposition.
BRANCHINGCounterfactual branches. “What if” analysis.
CYCLICALFuture constrains past. Feedback loops.
DIRECTORIALDramatic tension drives events. Narrative arcs.

​
19 Composable Mechanisms (M1–M19)

Each mechanism handles one aspect of social simulation and can be enabled/disabled independently:
  • Entity state tracking and lifecycle
  • Knowledge provenance and information flow
  • Dialog synthesis with voice consistency
  • Emotional dynamics and relationship evolution
  • Temporal reasoning and causal chains
  • Fidelity management (controls simulation depth vs. cost)

​
Templates

21 verified templates, 0.15–0.15–1.00 per run:
TemplateUse Case
board-meetingCorporate strategy simulation
castaway-colonySurvival group dynamics
mars-missionLong-duration isolation
first-contactCross-cultural negotiation
and 17 more…

​
Usage

git clone https://github.com/timepoint-ai/timepoint-pro.git
cd timepoint-pro
pip install -e .
export OPENROUTER_API_KEY="your-key"

python run.sh --template board-meeting --fidelity 3

​
Export Formats

  • Fountain — screenplay format
  • JSONL — training data for fine-tuning
  • SQLite — structured database export
  • TDF — Timepoint Data Format for cross-service interchange

​
Architecture

  • LLM: OpenRouter exclusively (fully standalone — no dependency on other Timepoint services)
  • Graph: LangGraph for orchestration, NetworkX for social graph
  • Build: Poetry + uv