How crescentschaos/cressyai works
CressyAI is a local-first GGUF chat application with a browser UI, server-side conversation and model handling, and optional research, memory, knowledge-base, and agent capabilities. The main workflow is a user sending a chat request through the UI to the server, which selects a model and may orchestrate tools before streaming a reply. The graph preserves the documented feature branches; some links are supported only by the README or SOURCE INDEX, since much of the server and UI source was not excerpted.
Main entry points: server.py Chat server · toolbox.py Tool orchestration · geo.py Places and weather · websearch.py Web search
Components 23
Chat application
- Browser UI
js - Chat server
server.py - Chat history
server.py
Models and agent
- Model manager
modelmgr.py - Agent mode
agent.py - Agent tools
agenttools.py - Tool orchestration
toolbox.py - Project files
fstools.py
Memory and knowledge
- Long-term memory
memory.py - Semantic embeddings
embed.py - Knowledge base
rag.py
Research tools
- Web search
websearch.py - Wikipedia lookup
wikipedia.py - GitHub lookup
github.py
Places and nature
- Animal lookup
animal.py - iNaturalist
inaturalist.py - Places and weather
geo.py - Geospatial math
geomath.py - Open-Meteo
openmeteo.py - OpenStreetMap
openstreetmap.py
Other components
- User
- SearXNG
- External APIs
Connections 32
- User → Browser UI: uses
- Browser UI → Chat server: sends requests
- Chat server → Browser UI: streams replies
- Chat server → Chat history: saves chats
server.py - Chat server → Model manager: loads models
server.py - Chat server → Tool orchestration: orchestrates tools
server.py - Chat server → Agent mode: runs agent
server.py - Agent mode → Agent tools: uses actions
agent.py - Agent tools → Project files: confines file access
- Tool orchestration → Project files: offers file tools
toolbox.py - Chat server → Long-term memory: uses memory
server.py - Long-term memory → Semantic embeddings: embeds for recall
memory.py - Long-term memory → Knowledge base: shares retrieval
memory.py - Chat server → Knowledge base: queries documents
server.py - Tool orchestration → Web search: searches web
- Tool orchestration → Wikipedia lookup: looks up articles
- Tool orchestration → GitHub lookup: looks up repositories
- Tool orchestration → iNaturalist: queries nature
- Tool orchestration → Places and weather: queries places
- Animal lookup → Web search: uses search
animal.py - Animal lookup → iNaturalist: uses species data
animal.py - Animal lookup → Wikipedia lookup: uses summaries
animal.py - GitHub lookup → Web search: uses web access
github.py - Places and weather → Geospatial math: uses calculations
geo.py - Places and weather → Open-Meteo: fetches weather
geo.py - Places and weather → OpenStreetMap: fetches places
geo.py - Open-Meteo → Geospatial math: validates coordinates
openmeteo.py - OpenStreetMap → Geospatial math: uses calculations
openstreetmap.py - Web search → SearXNG: searches through
- iNaturalist → External APIs: queries API
- Open-Meteo → External APIs: fetches forecasts
- OpenStreetMap → External APIs: queries map data