Skip to content

Self-improving AI Employees for everyone

docs: drop final baseChatModelFactory + Wave 6 Phase 3 references

trohitg699d998586fa
diff --git a/CLAUDE.md b/CLAUDE.md
index 17375c0c..9544d888 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -2001,7 +2001,7 @@ deploy_workflow() -> Sets up triggers, returns immediately
### AI Chat Model System (5-Layer Architecture)
- **Visual Components**: Circular node design with real-time status indicators (green/yellow/red)
-- **Node Definitions**: Factory pattern using `createBaseChatModel()` with provider configurations
+- **Node Definitions**: Backend NodeSpec (`server/nodes/models/<provider>.py`) is the single source of truth; the frontend renders via `lib/nodeSpec.ts` + `adapters/nodeSpecToDescription.ts`
- **Parameter System**: Universal renderer with drag-and-drop template variables (`{{variable}}`)
- **API Key Management**: Secure localStorage with base64 encryption and LangChain validation
- **Execution Engine**: Routes AI nodes to Python Flask backend with auto-injection of API keys
@@ -2152,42 +2152,8 @@ Extended thinking and reasoning capabilities for supported AI models. When enabl
| **Groq** | qwen3-32b | `reasoningFormat` ('parsed' or 'hidden') | format | 'parsed' returns reasoning, 'hidden' returns only final answer |
| **Cerebras** | qwen-3-235b | `reasoningFormat` ('parsed' or 'hidden') | format | Same format-based reasoning as Groq Qwen |
-#### Frontend Parameters (`client/src/factories/baseChatModelFactory.ts`)
-```typescript
-STANDARD_PARAMETERS = {
- thinkingEnabled: {
- displayName: 'Thinking/Reasoning Mode',
- name: 'thinkingEnabled',
- type: 'boolean',
- default: false,
- description: 'Enable extended thinking for supported models'
- },
- thinkingBudget: {
- displayName: 'Thinking Budget (Tokens)',
- name: 'thinkingBudget',
- type: 'number',
- default: 2048,
- typeOptions: { minValue: 1024, maxValue: 16000 },
- displayOptions: { show: { thinkingEnabled: [true] } }
- },
- reasoningEffort: {
- displayName: 'Reasoning Effort',
- name: 'reasoningEffort',
- type: 'options',
- options: ['minimal', 'low', 'medium', 'high'],
- default: 'medium',
- displayOptions: { show: { thinkingEnabled: [true] } }
- },
- reasoningFormat: {
- displayName: 'Reasoning Format',
- name: 'reasoningFormat',
- type: 'options',
- options: ['parsed', 'hidden'],
- default: 'parsed',
- displayOptions: { show: { thinkingEnabled: [true] } }
- }
-}
-```
+#### Thinking/Reasoning Parameters
+The thinking/reasoning fields (`thinkingEnabled`, `thinkingBudget`, `reasoningEffort`, `reasoningFormat`) live in the backend NodeSpec for each chat model (`server/nodes/models/<provider>.py`) and are surfaced through `AIChatModelParams` in `server/models/nodes.py`. The frontend renders them automatically via the universal parameter panel.
#### Backend Implementation (`server/services/ai.py`)
The AI service extracts thinking content from LangChain response objects:
diff --git a/client/vitest.config.ts b/client/vitest.config.ts
index 432e9989..a26f072f 100644
--- a/client/vitest.config.ts
+++ b/client/vitest.config.ts
@@ -19,7 +19,6 @@ export default defineConfig({
reporter: ['text', 'html'],
include: [
'src/adapters/**',
- 'src/factories/**',
'src/types/**',
'src/hooks/useApiKeys.ts',
'src/components/CredentialsModal.tsx',
diff --git a/server/models/node_metadata.py b/server/models/node_metadata.py
index d0715624..62b6db95 100644
--- a/server/models/node_metadata.py
+++ b/server/models/node_metadata.py
@@ -1,15 +1,15 @@
"""Per-node display metadata.
-Receives the UI-only fields (``displayName``, ``icon``, ``group``,
-``subtitle``, ``description``) that today live in the frontend
-``client/src/nodeDefinitions/*.ts`` files. Wave 6 Phase 3 migrates each
-node-definition file's display metadata into this dict; until then
-NodeSpec falls back to the node type id for missing entries.
-
-Single source of truth for everything the parameter panel + palette
-needs to render a node header. Kept as a plain dict (not Pydantic) for
-fast iteration during the migration — switch to a TypedDict / Pydantic
-model once the schema stabilises.
+Holds the UI-only fields (``displayName``, ``icon``, ``group``,
+``subtitle``, ``description``) for every node type. This dict is the
+single source of truth for everything the parameter panel + palette
+need to render a node header — the frontend consumes it via the
+NodeSpec endpoints (see ``server/routers/schemas.py`` and
+``server/nodes/README.md``). Missing entries fall back to the node
+type id.
+
+Kept as a plain dict (not Pydantic) for fast iteration — switch to a
+TypedDict / Pydantic model once the schema stabilises.
"""
from typing import Any, Literal, Optional, TypedDict
diff --git a/server/routers/schemas.py b/server/routers/schemas.py
index 5e652f86..0ca84bea 100644
--- a/server/routers/schemas.py
+++ b/server/routers/schemas.py
@@ -55,9 +55,9 @@ async def list_schemas():
async def get_node_input(node_type: str, response: Response):
"""Return the JSON Schema for a node type's input parameters.
- Wave 6 Phase 1: parallels the output endpoint above. 404 when no
- Pydantic model is registered for the type — frontend falls back to
- the legacy nodeDefinitions/* properties array until Phase 3 lands.
+ Parallels the output endpoint above. 404 when no Pydantic model is
+ registered for the type — the frontend's ``lib/nodeSpec.ts`` adapter
+ treats a missing input schema as an empty parameter set.
"""
schema = get_node_input_schema(node_type)