How fwdai/reticle works
Reticle is a local-first desktop workbench for designing, running, evaluating, and inspecting LLM scenarios and agents. Engineers configure a scenario or agent, execute it against provider models (and tools, where configured), inspect outputs and traces, and retain run history locally; eval datasets and assertions support repeatable checks. The graph preserves the README’s major product areas. Some links are based on documented capabilities and the file tree because their implementation was not sampled.
Main entry points: App.tsx App shell · storage Storage API · Scenarios Scenarios · Agents Agents
Components 23
Design and execution
- Scenarios
src/features/Scenarios - Scenario actions
src/actions/scenarioActions.ts - Agents
src/features/Agents - Agent actions
src/actions/agentActions.ts - Agent runtime
src/lib/agent - Human input
src/actions/agentHumanInput.ts - Tools
src/features/Tools - Templates
src/features/Templates
Evaluation and inspection
- Eval engine
src/lib/evals - Eval controls
src/components/evals - Run history
src/features/Runs - Trace visualizers
src/features/Runs/MainContent/RunDetail
Runtime integrations
- Tool execution
src/components/Tools/Detail/Code/useToolExecution.ts - Provider gateway
src/lib/gateway - Local proxy
src-tauri/src/server.rs - Code runner
src/lib/runner
Local services
- Storage API
src/lib/storage - SQLite database
src-tauri/src/database.rs
Other components
- Engineer
- Model providers
Connections 30
- Engineer → App shell: uses
- App shell → Dashboard: routes to
- App shell → Scenarios: routes to
- App shell → Agents: routes to
- App shell → Tools: routes to
- App shell → Run history: routes to
- App shell → Settings: routes to
- App shell → Templates: routes to
- Scenarios → Scenario actions: invokes
- Scenario actions → Provider gateway: runs model requests
- Agents → Agent actions: invokes
- Agent actions → Agent runtime: runs agent
- Agent runtime → Provider gateway: requests model
- Agent actions → Human input: handles prompts
- Tools → Tool execution: tests code tools
- Tool execution → Code runner: spawns scripts
- Eval controls → Eval engine: runs assertions
- Run history → Trace visualizers: renders traces
- Provider gateway → Local proxy: sends requests
- Local proxy → Model providers: forwards requests
- Local proxy → SQLite database: reads API keys
- Settings → Storage API: manages settings
- Dashboard → Storage API: loads summaries
- Scenarios → Storage API: stores scenarios
- Agents → Storage API: stores agents
- Tools → Storage API: stores tools
- Templates → Storage API: stores templates
- Run history → Storage API: loads history
- Eval engine → Storage API: stores eval data
- Storage API → SQLite database: reads and writes