RL Lab — interactive RL builder

Lives at /rl/lab in the AlphaSwarm operator UI (alphaswarm_client). Combines several surfaces:

Tab	Purpose	Component
Experiment (/rl/lab)	Compose env + reward + observation + action + agent + ensembler into one RLExperimentSpec on a drag-and-drop canvas, save, train.	RlLabRoute (uses WorkflowEditor + xyflow)
Reward / Observation / Agent / Experiment / Backbone / Advantage builders	Pick a registered class for the given rl_kind, fill its kwargs schema, POST a {class, module_path, kwargs} build-spec.	Generic RlBuilder.tsx, parametrised per route (there is no dedicated Environment-builder tab or component).
Component library	Browse every registered RL component, filter by tag / source / category.	RlComponentLibrary.tsx

Routes

Path	Component
/rl/lab	RlLabRoute
/rl/library	RlLibraryRoute (renders RlComponentLibrary)
/rl/builder/reward	RlRewardBuilderRoute (generic RlBuilder, kind="rl_reward")
/rl/builder/observation	RlObservationBuilderRoute (generic RlBuilder, kind="rl_observation")
/rl/builder/agent	RlAgentBuilderRoute (generic RlBuilder, kind="rl_agent")
/rl/builder/experiment	RlExperimentBuilderRoute (generic RlBuilder, kind="rl_experiment")
/rl/builder/backbone	RlBackboneBuilderRoute (generic RlBuilder, kind="rl_policy_backbone")
/rl/builder/advantage	RlAdvantageBuilderRoute (generic RlBuilder, kind="rl_advantage_estimator")
/rl/runs	RlRunsRoute (RlRunsPage)
/rl/runs/[id]	RlRunDetailRoute
/rl/runs/[id]/replay	RlReplayRoute (RlReplayViewer)
/rl	RlHomeRoute (quick-train, application registry browser).
/rl/zoo	RlZooRoute — RL agent zoo (RlZooPage).

There is no /rl/builder/env route or EnvironmentBuilder component; env composition happens inside the Experiment canvas above. The Experiment canvas uses the existing WorkflowEditor + @xyflow/react stack. The serializer in alphaswarm_client/src/components/rl/rlSerializer.ts turns a FlowGraph into an RLExperimentSpec payload by bucketising nodes via their palette group (env / observation / action / reward / termination / agent / data pipeline / ensembler).

API surface used

The lab calls the API endpoints in alphaswarm_rl/src/alphaswarm_rl/api/routes/rl.py:

• GET /rl/components — kind counts.
• GET /rl/components/{kind} — list registered components per kind.
• POST /rl/lab/preview-reward — reward decomposition.
• POST /rl/lab/preview-observation — observation shape + features.
• POST /rl/lab/preview-action — action transform sample.
• POST /rl/specs — persist a spec.
• POST /rl/specs/{slug}/run — kick off train / evaluate / paper / replay / walk-forward via the matching Celery task.
• GET /rl/runs / GET /rl/runs/{id} / .../equity / .../trajectories / .../reward-decomposition / .../episodes / .../actions — runs ledger + step-level data served from DuckDB views over the Iceberg tables.
• POST /rl/runs/{id}/replay — re-roll a saved policy on a new window.
• POST /rl/data-pipelines/preview — show first rows + array shapes.

Run replay

The replay viewer (/rl/runs/[id]/replay) loads:

• rl.equity_curves rows for the chosen episode (slider populates from the row count).
• rl.trajectories rows for the chosen episode (each step shows reward + info JSON).

Both come from the DuckDB views generated by alphaswarm_rl/src/alphaswarm_rl/trajectories/duckdb_views.py.