Donna
1e2073a81c
Add edge_mappings table for AI-assisted field mapping
...
New table stores:
- Edge identification (from_node → to_node)
- Analysis results (confidence, level, method)
- Proposed mapping (AI-generated)
- User mapping (overrides)
- Confirmation status
Indexes:
- By flow_id for listing
- Unique on (flow_id, from_node, to_node) for upsert
This supports the edge analysis API for visual wiring in the canvas.
Co-authored-by: Dan
2026-01-26 07:23:03 +00:00
Donna
a6f44b1fc9
Set up Alembic database migrations
...
- Initialize Alembic in bloxserver/
- Configure for async SQLAlchemy (converts async URLs to sync for migrations)
- Generate initial migration with all existing tables:
- users (Clerk sync, Stripe billing)
- flows (workflow definitions)
- triggers (webhook, schedule, manual, event)
- executions (run history)
- user_api_keys (BYOK encrypted storage)
- usage_records (billing metrics)
- stripe_events (webhook idempotency)
Usage:
cd bloxserver
alembic upgrade head # Apply migrations
alembic revision --autogenerate -m 'description' # New migration
Co-authored-by: Dan
2026-01-26 07:21:56 +00:00
dullfig
a69eae79c5
Add FlowRunner - execution bridge for BloxServer flows
...
FlowRunner connects Flow domain model to xml-pipeline's StreamPump:
- Lifecycle management: start(), stop() with state tracking
- Flow → OrganismConfig conversion: nodes become listeners
- Trigger handling: webhook, schedule, manual triggers
- Message injection: inject() to specific nodes, trigger() by trigger ID
- Execution tracking: events log, status reporting, message counting
States: CREATED → STARTING → RUNNING → STOPPING → STOPPED (or ERROR)
Usage:
runner = FlowRunner(flow)
await runner.start()
thread_id = await runner.trigger("webhook_id", payload)
await runner.stop()
Known limitation: Tool nodes need adapter wrappers to convert
@tool decorator (ToolResult) to handler pattern (HandlerResponse).
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-22 22:20:59 -08:00
dullfig
9ba77b843d
Add Flow domain model for BloxServer
...
Domain model bridges frontend canvas, database, and xml-pipeline:
- nodes.py: AgentNode, ToolNode, GatewayNode with serialization
- Built-in tool mappings (calculate, fetch, shell, etc.)
- Agent config (prompt, model, temperature)
- Gateway config (federation + REST API)
- edges.py: Edge connections with conditions
- Auto-compute peers from edges
- Cycle detection, entry/exit node finding
- triggers.py: Webhook, Schedule, Manual, Event triggers
- Config dataclasses for each type
- Factory functions for common patterns
- flow.py: Main Flow class aggregating all components
- to_organism_yaml(): Generate xml-pipeline config
- from_canvas_json() / to_canvas_json(): React Flow compat
- validate(): Check for errors before execution
- to_db_dict(): Database serialization
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-22 22:15:51 -08:00
dullfig
a5c00c1e90
Add BloxServer API scaffold + architecture docs
...
BloxServer API (FastAPI + SQLAlchemy async):
- Database models: users, flows, triggers, executions, usage tracking
- Clerk JWT auth with dev mode bypass for local testing
- SQLite support for local dev, PostgreSQL for production
- CRUD routes for flows, triggers, executions
- Public webhook endpoint with token auth
- Health/readiness endpoints
- Pydantic schemas with camelCase aliases for frontend
- Docker + docker-compose setup
Architecture documentation:
- Librarian architecture with RLM-powered query engine
- Stripe billing integration (usage-based, trials, webhooks)
- LLM abstraction layer (rate limiting, semantic cache, failover)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-22 22:04:25 -08:00
