inputlayer-client-dev 0.1.0__tar.gz → 0.1.0.dev913__tar.gz

inputlayer_client_dev-0.1.0.dev913/.gitignore

@@ -0,0 +1 @@
+.coverage

inputlayer_client_dev-0.1.0/README.md → inputlayer_client_dev-0.1.0.dev913/PKG-INFO

@@ -1,6 +1,39 @@
+Metadata-Version: 2.4
+Name: inputlayer-client-dev
+Version: 0.1.0.dev913
+Summary: Python Object-Logic Mapper for InputLayer knowledge graph engine
+License-Expression: Apache-2.0
+Requires-Python: >=3.10
+Requires-Dist: pydantic<3,>=2.0
+Requires-Dist: websockets<15,>=12.0
+Provides-Extra: all
+Requires-Dist: langchain-core<2,>=0.3; extra == 'all'
+Requires-Dist: langgraph<2,>=0.2; extra == 'all'
+Requires-Dist: pandas>=1.5; extra == 'all'
+Requires-Dist: typing-extensions>=4.0; (python_version < '3.11') and extra == 'all'
+Provides-Extra: dev
+Requires-Dist: langchain-core<2,>=0.3; extra == 'dev'
+Requires-Dist: langchain-openai>=1.0; extra == 'dev'
+Requires-Dist: langgraph<2,>=0.2; extra == 'dev'
+Requires-Dist: mypy>=1.13; extra == 'dev'
+Requires-Dist: pandas>=1.5; extra == 'dev'
+Requires-Dist: pytest-asyncio>=0.21; extra == 'dev'
+Requires-Dist: pytest>=7.0; extra == 'dev'
+Requires-Dist: ruff>=0.8; extra == 'dev'
+Requires-Dist: typing-extensions>=4.0; (python_version < '3.11') and extra == 'dev'
+Provides-Extra: langchain
+Requires-Dist: langchain-core<2,>=0.3; extra == 'langchain'
+Provides-Extra: langgraph
+Requires-Dist: langchain-core<2,>=0.3; extra == 'langgraph'
+Requires-Dist: langgraph<2,>=0.2; extra == 'langgraph'
+Requires-Dist: typing-extensions>=4.0; (python_version < '3.11') and extra == 'langgraph'
+Provides-Extra: pandas
+Requires-Dist: pandas>=1.5; extra == 'pandas'
+Description-Content-Type: text/markdown
+
 # inputlayer-client-dev
 
-Python Object-Logic Mapper (OLM) for [InputLayer](https://github.com/inputlayer/inputlayer) - the symbolic reasoning engine for AI agents.
+Python Object-Logic Mapper (OLM) for [InputLayer](https://github.com/inputlayer/inputlayer) - the streaming reasoning layer for AI systems.
 
 Write Python. No query syntax required. The OLM compiles typed Python classes into InputLayer queries over WebSocket.
 
@@ -9,8 +42,10 @@ Write Python. No query syntax required. The OLM compiles typed Python classes in
 ```bash
 pip install inputlayer-client-dev
 
-# With pandas DataFrame support
-pip install inputlayer-client-dev[pandas]
+# With extras
+pip install inputlayer-client-dev[pandas]      # DataFrame support
+pip install inputlayer-client-dev[langchain]   # LangChain integration
+pip install inputlayer-client-dev[all]         # everything
 ```
 
 Requirements: Python 3.10+ and a running InputLayer server.
@@ -180,6 +215,195 @@ def on_update(event):
     print(f"{event.count} new readings")
 ```
 
+## LangChain Integration
+
+```bash
+pip install inputlayer-client-dev[langchain]
+```
+
+The full integration guide lives at [docs/guides/langchain](../../docs/content/docs/guides/langchain.mdx). Highlights:
+
+### Vector store
+
+Drop-in `langchain_core.vectorstores.VectorStore` backed by an InputLayer `Relation`. Embeds documents through any LangChain `Embeddings` instance, supports metadata filters, deletion, and `as_retriever()`:
+
+```python
+from inputlayer import Relation, Vector
+from inputlayer.integrations.langchain import InputLayerVectorStore
+from langchain_openai import OpenAIEmbeddings
+
+class Chunk(Relation):
+    id: str
+    content: str
+    source: str
+    embedding: Vector
+
+await kg.define(Chunk)
+vs = InputLayerVectorStore(kg=kg, relation=Chunk, embeddings=OpenAIEmbeddings())
+await vs.aadd_texts(["..."], metadatas=[{"source": "wiki"}], ids=["doc1"])
+docs = await vs.asimilarity_search("query", k=5, filter={"source": "wiki"})
+```
+
+### Retriever
+
+`InputLayerRetriever` runs in vector mode (with an `Embeddings` instance) or in InputLayer Query Language mode with safe `:input` parameter binding:
+
+```python
+from inputlayer.integrations.langchain import InputLayerRetriever
+
+retriever = InputLayerRetriever(
+    kg=kg,
+    query="?article(I, T, C, Cat, E), user_interest(:input, Cat)",
+    page_content_columns=["content"],
+    metadata_columns=["title", "category"],
+)
+result = await retriever.ainvoke("alice")
+```
+
+### Structured agent tools
+
+`tools_from_relations` generates one `StructuredTool` per `Relation` with typed equality, range, and IN-list filters. The LLM never has to write IQL:
+
+```python
+from inputlayer.integrations.langchain import tools_from_relations
+tools = tools_from_relations(kg, [Employee, Article])
+agent = create_tool_calling_agent(llm, tools, prompt)
+```
+
+For agents that genuinely need raw IQL access, `InputLayerIQLTool` is the escape hatch. All components support both sync (`invoke`) and async (`ainvoke`) and are safe to use inside Jupyter, FastAPI, and LangGraph.
+
+### Examples
+
+See [`examples/langchain/`](examples/langchain/) — 17 examples covering AI/LLM integration patterns:
+
+```bash
+# List all examples
+uv run python -m examples.langchain.runner --list
+
+# Run specific examples
+uv run python -m examples.langchain.runner 1 3 9
+
+# Run a range
+uv run python -m examples.langchain.runner 1-5
+
+# Run all
+uv run python -m examples.langchain.runner --all
+```
+
+| # | Example | Description |
+|---|---------|-------------|
+| 1 | Retriever + IQL | Join queries with `{input}` placeholder |
+| 2 | Vector search | Cosine similarity with distance filter |
+| 3 | Tool for agents | Raw IQL + template mode |
+| 4 | LCEL chain | Full retriever \| prompt \| llm \| parser pipeline |
+| 5 | KG building | Extract facts from documents with LLM |
+| 6 | Explainable RAG | `.why()` proof trees + `.why_not()` explanations |
+| 7 | Multi-hop reasoning | Transitive closure over org graph |
+| 8 | Conversational memory | Chat turns as facts, rules derive context |
+| 9 | Access-controlled RAG | Clearance-based document filtering via rules |
+| 10 | Multi-agent | Researcher + fact-checker with shared KG |
+| 11 | Anomaly detection | Salary band rules flag violations |
+| 12 | Hallucination detection | Ground LLM claims against KG facts |
+| 13 | Guardrails | Policy rules block unsafe content |
+| 14 | GraphRAG | Entity extraction + community detection |
+| 15 | Semantic caching | Cache LLM responses, topic-based matching |
+| 16 | Recommendation engine | Collaborative filtering via IQL rules |
+| 17 | Data lineage | Source attribution + conflict detection |
+
+Requires a running InputLayer server and optionally LM Studio (or any OpenAI-compatible server) for LLM examples.
+
+## LangGraph Integration
+
+Install the langgraph extra:
+
+```bash
+pip install inputlayer-client-dev[langgraph]
+```
+
+The LangGraph integration provides:
+
+- **`InputLayerCheckpointer`**: Persist graph state in an InputLayer KG. Supports `prune_thread()` / `adelete_thread()` for storage management and full async/sync parity.
+- **`InputLayerMemory`**: Semantic long-term memory. Stores conversation turns as facts, derives active topics and relevant context via rules. Supports `adelete_thread()` for thread cleanup.
+- **`kg_node`**: Factory for query/insert/delete graph nodes.
+- **`kg_router`**: Conditional edge routing driven by IQL queries.
+- **`InputLayerState`**: TypedDict base class with the required `kg` field for graph state.
+- **`escape_iql`**: String escaping for safe IQL interpolation in parameterized queries.
+
+```python
+from inputlayer import InputLayer
+from inputlayer.integrations.langgraph import (
+    InputLayerCheckpointer,
+    InputLayerMemory,
+    InputLayerState,
+    escape_iql,
+    kg_node,
+    kg_router,
+)
+from langgraph.graph import END, StateGraph
+
+class MyState(InputLayerState):
+    question: str
+    answer: str
+
+async with InputLayer("ws://localhost:8080/ws", username="admin", password="...") as il:
+    kg = il.knowledge_graph("my_agent")
+
+    # Checkpointer: persist graph state across process restarts
+    checkpointer = InputLayerCheckpointer(kg=kg)
+    await checkpointer.setup()
+
+    # Memory: semantic recall with rule-derived context
+    memory = InputLayerMemory(kg=kg)
+    await memory.setup()
+
+    # Build a graph with KG-driven nodes and routing
+    graph = StateGraph(MyState)
+    graph.add_node("search", kg_node(query="?relevant(X, Y)", state_key="results"))
+    graph.add_node("recall", memory.recall_node(state_key="context"))
+    graph.add_node("store", memory.store_node(state_key="new_message"))
+    graph.set_entry_point("recall")
+    graph.add_edge("recall", "search")
+    graph.add_edge("search", "store")
+    graph.add_edge("store", END)
+
+    app = graph.compile(checkpointer=checkpointer)
+```
+
+### LangGraph Examples
+
+See [`examples/langgraph/`](examples/langgraph/) - 12 examples covering agent patterns:
+
+```bash
+# List all examples
+uv run python -m examples.langgraph.runner --list
+
+# Run specific examples
+uv run python -m examples.langgraph.runner 1 10 11
+
+# Run a range
+uv run python -m examples.langgraph.runner 1-5
+
+# Run all
+uv run python -m examples.langgraph.runner --all
+```
+
+| # | Example | Description |
+|---|---------|-------------|
+| 1 | Reasoning loop | Accumulate facts, rules decide when to stop |
+| 2 | Investigation | Multi-step evidence gathering |
+| 3 | Human-in-the-loop | Policy rules gate actions for approval |
+| 4 | Branching pipeline | Route documents through parallel analysis |
+| 5 | Self-correcting agent | Validation rules catch and fix errors |
+| 6 | Collaborative planning | Multi-agent task decomposition |
+| 7 | Event correlation | Pattern detection across event streams |
+| 8 | Tool selection | Rules pick the right tool per context |
+| 9 | Streaming aggregation | Threshold-based alerts from streaming data |
+| 10 | Resumable graph | Checkpoint, crash, resume from persisted state |
+| 11 | Semantic memory | Store turns as facts, recall derived context |
+| 12 | Resumable chat | Checkpointer and memory together on one KG |
+
+Requires a running InputLayer server. Examples marked [LLM] need LM Studio (or any OpenAI-compatible server) at `localhost:1234`. Set `INPUTLAYER_URL`, `INPUTLAYER_USER`, `INPUTLAYER_PASSWORD` to override server defaults.
+
 ## Sync Client
 
 For scripts, notebooks, and non-async contexts:
@@ -260,7 +484,7 @@ The autodetector diffs your current Python models against the last migration's s
 | `revoke_access(username)` | Revoke per-KG access |
 | `list_acl()` | List access control entries |
 | `explain(*select, ...)` | Show query plan without executing |
-| `execute(datalog)` | Execute raw Datalog |
+| `execute(iql)` | Execute raw IQL |
 | `status()` | Get server status |
 | `compact()` | Trigger storage compaction |
 
@@ -334,15 +558,18 @@ See the [`examples/`](examples/) directory:
 | `08_session_rules.py` | Ad-hoc ephemeral views |
 | `09_access_control.py` | User/ACL management |
 | `10_migrations.py` | Django-style schema versioning |
+| `langchain/` | 17 LangChain integration examples (see above) |
 
 ## Development
 
 ```bash
 cd packages/inputlayer-py
-pip install -e ".[dev]"
-python -m pytest tests/ -v
+uv sync --extra dev
+uv run pytest tests/ -v
+uv run ruff check src/ tests/
+uv run mypy src/inputlayer/
 ```
 
 ## License
 
-AGPL-3.0
+Apache 2.0 + Commons Clause. See [LICENSE](../../LICENSE).

inputlayer_client_dev-0.1.0/PKG-INFO → inputlayer_client_dev-0.1.0.dev913/README.md

@@ -1,22 +1,6 @@
-Metadata-Version: 2.4
-Name: inputlayer-client-dev
-Version: 0.1.0
-Summary: Python Object-Logic Mapper for InputLayer knowledge graph engine
-License-Expression: AGPL-3.0
-Requires-Python: >=3.10
-Requires-Dist: pydantic>=2.0
-Requires-Dist: websockets>=12.0
-Provides-Extra: dev
-Requires-Dist: pandas>=1.5; extra == 'dev'
-Requires-Dist: pytest-asyncio>=0.21; extra == 'dev'
-Requires-Dist: pytest>=7.0; extra == 'dev'
-Provides-Extra: pandas
-Requires-Dist: pandas>=1.5; extra == 'pandas'
-Description-Content-Type: text/markdown
-
 # inputlayer-client-dev
 
-Python Object-Logic Mapper (OLM) for [InputLayer](https://github.com/inputlayer/inputlayer) - the symbolic reasoning engine for AI agents.
+Python Object-Logic Mapper (OLM) for [InputLayer](https://github.com/inputlayer/inputlayer) - the streaming reasoning layer for AI systems.
 
 Write Python. No query syntax required. The OLM compiles typed Python classes into InputLayer queries over WebSocket.
 
@@ -25,8 +9,10 @@ Write Python. No query syntax required. The OLM compiles typed Python classes in
 ```bash
 pip install inputlayer-client-dev
 
-# With pandas DataFrame support
-pip install inputlayer-client-dev[pandas]
+# With extras
+pip install inputlayer-client-dev[pandas]      # DataFrame support
+pip install inputlayer-client-dev[langchain]   # LangChain integration
+pip install inputlayer-client-dev[all]         # everything
 ```
 
 Requirements: Python 3.10+ and a running InputLayer server.
@@ -196,6 +182,195 @@ def on_update(event):
     print(f"{event.count} new readings")
 ```
 
+## LangChain Integration
+
+```bash
+pip install inputlayer-client-dev[langchain]
+```
+
+The full integration guide lives at [docs/guides/langchain](../../docs/content/docs/guides/langchain.mdx). Highlights:
+
+### Vector store
+
+Drop-in `langchain_core.vectorstores.VectorStore` backed by an InputLayer `Relation`. Embeds documents through any LangChain `Embeddings` instance, supports metadata filters, deletion, and `as_retriever()`:
+
+```python
+from inputlayer import Relation, Vector
+from inputlayer.integrations.langchain import InputLayerVectorStore
+from langchain_openai import OpenAIEmbeddings
+
+class Chunk(Relation):
+    id: str
+    content: str
+    source: str
+    embedding: Vector
+
+await kg.define(Chunk)
+vs = InputLayerVectorStore(kg=kg, relation=Chunk, embeddings=OpenAIEmbeddings())
+await vs.aadd_texts(["..."], metadatas=[{"source": "wiki"}], ids=["doc1"])
+docs = await vs.asimilarity_search("query", k=5, filter={"source": "wiki"})
+```
+
+### Retriever
+
+`InputLayerRetriever` runs in vector mode (with an `Embeddings` instance) or in InputLayer Query Language mode with safe `:input` parameter binding:
+
+```python
+from inputlayer.integrations.langchain import InputLayerRetriever
+
+retriever = InputLayerRetriever(
+    kg=kg,
+    query="?article(I, T, C, Cat, E), user_interest(:input, Cat)",
+    page_content_columns=["content"],
+    metadata_columns=["title", "category"],
+)
+result = await retriever.ainvoke("alice")
+```
+
+### Structured agent tools
+
+`tools_from_relations` generates one `StructuredTool` per `Relation` with typed equality, range, and IN-list filters. The LLM never has to write IQL:
+
+```python
+from inputlayer.integrations.langchain import tools_from_relations
+tools = tools_from_relations(kg, [Employee, Article])
+agent = create_tool_calling_agent(llm, tools, prompt)
+```
+
+For agents that genuinely need raw IQL access, `InputLayerIQLTool` is the escape hatch. All components support both sync (`invoke`) and async (`ainvoke`) and are safe to use inside Jupyter, FastAPI, and LangGraph.
+
+### Examples
+
+See [`examples/langchain/`](examples/langchain/) — 17 examples covering AI/LLM integration patterns:
+
+```bash
+# List all examples
+uv run python -m examples.langchain.runner --list
+
+# Run specific examples
+uv run python -m examples.langchain.runner 1 3 9
+
+# Run a range
+uv run python -m examples.langchain.runner 1-5
+
+# Run all
+uv run python -m examples.langchain.runner --all
+```
+
+| # | Example | Description |
+|---|---------|-------------|
+| 1 | Retriever + IQL | Join queries with `{input}` placeholder |
+| 2 | Vector search | Cosine similarity with distance filter |
+| 3 | Tool for agents | Raw IQL + template mode |
+| 4 | LCEL chain | Full retriever \| prompt \| llm \| parser pipeline |
+| 5 | KG building | Extract facts from documents with LLM |
+| 6 | Explainable RAG | `.why()` proof trees + `.why_not()` explanations |
+| 7 | Multi-hop reasoning | Transitive closure over org graph |
+| 8 | Conversational memory | Chat turns as facts, rules derive context |
+| 9 | Access-controlled RAG | Clearance-based document filtering via rules |
+| 10 | Multi-agent | Researcher + fact-checker with shared KG |
+| 11 | Anomaly detection | Salary band rules flag violations |
+| 12 | Hallucination detection | Ground LLM claims against KG facts |
+| 13 | Guardrails | Policy rules block unsafe content |
+| 14 | GraphRAG | Entity extraction + community detection |
+| 15 | Semantic caching | Cache LLM responses, topic-based matching |
+| 16 | Recommendation engine | Collaborative filtering via IQL rules |
+| 17 | Data lineage | Source attribution + conflict detection |
+
+Requires a running InputLayer server and optionally LM Studio (or any OpenAI-compatible server) for LLM examples.
+
+## LangGraph Integration
+
+Install the langgraph extra:
+
+```bash
+pip install inputlayer-client-dev[langgraph]
+```
+
+The LangGraph integration provides:
+
+- **`InputLayerCheckpointer`**: Persist graph state in an InputLayer KG. Supports `prune_thread()` / `adelete_thread()` for storage management and full async/sync parity.
+- **`InputLayerMemory`**: Semantic long-term memory. Stores conversation turns as facts, derives active topics and relevant context via rules. Supports `adelete_thread()` for thread cleanup.
+- **`kg_node`**: Factory for query/insert/delete graph nodes.
+- **`kg_router`**: Conditional edge routing driven by IQL queries.
+- **`InputLayerState`**: TypedDict base class with the required `kg` field for graph state.
+- **`escape_iql`**: String escaping for safe IQL interpolation in parameterized queries.
+
+```python
+from inputlayer import InputLayer
+from inputlayer.integrations.langgraph import (
+    InputLayerCheckpointer,
+    InputLayerMemory,
+    InputLayerState,
+    escape_iql,
+    kg_node,
+    kg_router,
+)
+from langgraph.graph import END, StateGraph
+
+class MyState(InputLayerState):
+    question: str
+    answer: str
+
+async with InputLayer("ws://localhost:8080/ws", username="admin", password="...") as il:
+    kg = il.knowledge_graph("my_agent")
+
+    # Checkpointer: persist graph state across process restarts
+    checkpointer = InputLayerCheckpointer(kg=kg)
+    await checkpointer.setup()
+
+    # Memory: semantic recall with rule-derived context
+    memory = InputLayerMemory(kg=kg)
+    await memory.setup()
+
+    # Build a graph with KG-driven nodes and routing
+    graph = StateGraph(MyState)
+    graph.add_node("search", kg_node(query="?relevant(X, Y)", state_key="results"))
+    graph.add_node("recall", memory.recall_node(state_key="context"))
+    graph.add_node("store", memory.store_node(state_key="new_message"))
+    graph.set_entry_point("recall")
+    graph.add_edge("recall", "search")
+    graph.add_edge("search", "store")
+    graph.add_edge("store", END)
+
+    app = graph.compile(checkpointer=checkpointer)
+```
+
+### LangGraph Examples
+
+See [`examples/langgraph/`](examples/langgraph/) - 12 examples covering agent patterns:
+
+```bash
+# List all examples
+uv run python -m examples.langgraph.runner --list
+
+# Run specific examples
+uv run python -m examples.langgraph.runner 1 10 11
+
+# Run a range
+uv run python -m examples.langgraph.runner 1-5
+
+# Run all
+uv run python -m examples.langgraph.runner --all
+```
+
+| # | Example | Description |
+|---|---------|-------------|
+| 1 | Reasoning loop | Accumulate facts, rules decide when to stop |
+| 2 | Investigation | Multi-step evidence gathering |
+| 3 | Human-in-the-loop | Policy rules gate actions for approval |
+| 4 | Branching pipeline | Route documents through parallel analysis |
+| 5 | Self-correcting agent | Validation rules catch and fix errors |
+| 6 | Collaborative planning | Multi-agent task decomposition |
+| 7 | Event correlation | Pattern detection across event streams |
+| 8 | Tool selection | Rules pick the right tool per context |
+| 9 | Streaming aggregation | Threshold-based alerts from streaming data |
+| 10 | Resumable graph | Checkpoint, crash, resume from persisted state |
+| 11 | Semantic memory | Store turns as facts, recall derived context |
+| 12 | Resumable chat | Checkpointer and memory together on one KG |
+
+Requires a running InputLayer server. Examples marked [LLM] need LM Studio (or any OpenAI-compatible server) at `localhost:1234`. Set `INPUTLAYER_URL`, `INPUTLAYER_USER`, `INPUTLAYER_PASSWORD` to override server defaults.
+
 ## Sync Client
 
 For scripts, notebooks, and non-async contexts:
@@ -276,7 +451,7 @@ The autodetector diffs your current Python models against the last migration's s
 | `revoke_access(username)` | Revoke per-KG access |
 | `list_acl()` | List access control entries |
 | `explain(*select, ...)` | Show query plan without executing |
-| `execute(datalog)` | Execute raw Datalog |
+| `execute(iql)` | Execute raw IQL |
 | `status()` | Get server status |
 | `compact()` | Trigger storage compaction |
 
@@ -350,15 +525,18 @@ See the [`examples/`](examples/) directory:
 | `08_session_rules.py` | Ad-hoc ephemeral views |
 | `09_access_control.py` | User/ACL management |
 | `10_migrations.py` | Django-style schema versioning |
+| `langchain/` | 17 LangChain integration examples (see above) |
 
 ## Development
 
 ```bash
 cd packages/inputlayer-py
-pip install -e ".[dev]"
-python -m pytest tests/ -v
+uv sync --extra dev
+uv run pytest tests/ -v
+uv run ruff check src/ tests/
+uv run mypy src/inputlayer/
 ```
 
 ## License
 
-AGPL-3.0
+Apache 2.0 + Commons Clause. See [LICENSE](../../LICENSE).

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/01_quickstart.py

@@ -1,6 +1,8 @@
 """Quickstart: connect, define, insert, query."""
 
 import asyncio
+import os
+
 from inputlayer import InputLayer, Relation
 
 
@@ -13,7 +15,11 @@ class Employee(Relation):
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("quickstart")
 
         # Define schema

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/02_social_network.py

@@ -1,6 +1,7 @@
 """Social network: graph traversal, transitive closure, mutual follows."""
 
 import asyncio
+import os
 from typing import ClassVar
 
 from inputlayer import Derived, From, InputLayer, Relation
@@ -29,7 +30,11 @@ Reachable.rules = [
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("social")
 
         await kg.define(Follow)

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/03_rag_pipeline.py

@@ -1,6 +1,7 @@
 """RAG pipeline: vector + structured hybrid search."""
 
 import asyncio
+import os
 
 from inputlayer import HnswIndex, InputLayer, Relation, Vector
 
@@ -13,7 +14,11 @@ class Document(Relation):
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("rag")
 
         await kg.define(Document)

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/04_ecommerce.py

@@ -1,6 +1,7 @@
 """E-commerce: collaborative filtering, revenue aggregation."""
 
 import asyncio
+import os
 
 from inputlayer import InputLayer, Relation, count, sum_
 
@@ -12,7 +13,11 @@ class Purchase(Relation):
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("ecommerce")
 
         await kg.define(Purchase)

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/05_rbac.py

@@ -1,6 +1,7 @@
 """RBAC: transitive role inheritance."""
 
 import asyncio
+import os
 from typing import ClassVar
 
 from inputlayer import Derived, From, InputLayer, Relation
@@ -34,7 +35,11 @@ EffectiveRole.rules = [
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("rbac")
 
         await kg.define(RoleHierarchy, UserRole)

{inputlayer_client_dev-0.1.0 → inputlayer_client_dev-0.1.0.dev913}/examples/06_realtime_dashboard.py

@@ -1,8 +1,9 @@
 """Realtime dashboard: notifications + aggregation."""
 
 import asyncio
+import os
 
-from inputlayer import InputLayer, Relation, count, sum_
+from inputlayer import InputLayer, Relation, count
 
 
 class SensorReading(Relation):
@@ -12,7 +13,11 @@ class SensorReading(Relation):
 
 
 async def main():
-    async with InputLayer("ws://localhost:8080/ws", username="admin", password="admin") as il:
+    async with InputLayer(
+        os.environ.get("INPUTLAYER_URL", "ws://localhost:8080/ws"),
+        username=os.environ.get("INPUTLAYER_USER", "admin"),
+        password=os.environ.get("INPUTLAYER_PASSWORD", "admin"),
+    ) as il:
         kg = il.knowledge_graph("dashboard")
 
         # Register notification handler