Name change pipeline

AGENTS.md

@@ -22,7 +22,7 @@ A jambonz application controls phone calls by returning **arrays of verbs** —
 - **Webhook (HTTP)**: Your server receives POST requests and returns JSON verb arrays. Stateless and simple.
 - **WebSocket**: Persistent bidirectional connection. Required for real-time LLM agents, audio streaming, and TTS token streaming.
 
-**IMPORTANT**: Any application that uses a speech-to-speech verb (`openai_s2s`, `google_s2s`, `deepgram_s2s`, `ultravox_s2s`, `elevenlabs_s2s`, `s2s`, or `pipeline`) MUST use WebSocket transport.
+**IMPORTANT**: Any application that uses a speech-to-speech verb (`openai_s2s`, `google_s2s`, `deepgram_s2s`, `ultravox_s2s`, `elevenlabs_s2s`, `s2s`, or `agent`) MUST use WebSocket transport.
 
 ## Core Verbs
 
@@ -34,7 +34,7 @@ A jambonz application controls phone calls by returning **arrays of verbs** —
 ### AI & Real-time
 - **openai_s2s** / **google_s2s** / **deepgram_s2s** / **ultravox_s2s** / **elevenlabs_s2s** — Vendor-specific LLM voice conversation.
 - **s2s** — Generic LLM voice conversation (use when vendor is determined at runtime).
-- **pipeline** — Higher-level voice AI pipeline with integrated turn detection.
+- **agent** — Higher-level voice AI agent with integrated turn detection.
 - **dialogflow** — Google Dialogflow agent.
 - **stream** — Stream raw audio to a websocket endpoint.
 - **transcribe** — Real-time call transcription.
@@ -258,7 +258,7 @@ await client.calls.whisper(call_sid, {"verb": "say", "text": "Hello"})
 await client.calls.mute(call_sid, "mute")
 await client.calls.redirect(call_sid, "https://example.com/new")
 await client.calls.update(call_sid, {"call_status": "completed"})
-await client.calls.update_pipeline(call_sid, {"type": "update_instructions", "instructions": "New prompt"})
+await client.calls.update_agent(call_sid, {"type": "update_instructions", "instructions": "New prompt"})
 ```
 
 ## TTS Token Streaming
@@ -273,18 +273,18 @@ await session.flush_tts_tokens()
 await session.clear_tts_tokens()
 ```
 
-## Pipeline Updates
+## Agent Updates
 
-Update a running pipeline mid-conversation:
+Update a running agent mid-conversation:
 
 ```python
-await session.update_pipeline({"type": "update_instructions", "instructions": "Now help with billing."})
-await session.update_pipeline({"type": "inject_context", "messages": [{"role": "system", "content": "Customer is Gold tier."}]})
-await session.update_pipeline({"type": "update_tools", "tools": [...]})
-await session.update_pipeline({"type": "generate_reply", "user_input": "Override", "interrupt": True})
+await session.update_agent({"type": "update_instructions", "instructions": "Now help with billing."})
+await session.update_agent({"type": "inject_context", "messages": [{"role": "system", "content": "Customer is Gold tier."}]})
+await session.update_agent({"type": "update_tools", "tools": [...]})
+await session.update_agent({"type": "generate_reply", "user_input": "Override", "interrupt": True})
 ```
 
-## Tool Output (Pipeline)
+## Tool Output (Agent)
 
 When the LLM requests a tool call, return the result:
 
@@ -348,7 +348,7 @@ audio_svc.on("connection", on_audio_connection)
 | `llm:tool-output` | Tool call result (`tool_output()`) |
 | `tts:tokens` | Stream TTS text (`send_tts_tokens()`) |
 | `tts:flush` | End TTS stream (`flush_tts_tokens()`) |
-| `pipeline:update` | Pipeline update (`update_pipeline()`) |
+| `agent:update` | Agent update (`update_agent()`) |
 
 ## Common Patterns
 
@@ -361,17 +361,17 @@ jambonz.say(text="Welcome.").gather(
 ).say(text="No input. Goodbye.").hangup()
 ```
 
-### Voice Agent (Pipeline)
+### Voice Agent
 ```python
-session.pipeline(
+session.agent(
     stt={"vendor": "deepgram", "language": "en-US"},
     tts={"vendor": "cartesia", "voice": "sonic-english"},
     llm={"vendor": "openai", "model": "gpt-4o", "llmOptions": {
         "messages": [{"role": "system", "content": "You are a helpful assistant."}]
     }},
     turnDetection="krisp",
     bargeIn={"enable": True},
-    actionHook="/pipeline-done",
+    actionHook="/agent-done",
     eventHook="/events",
     toolHook="/tools",
 )
@@ -410,9 +410,9 @@ jambonz.say(text="Connecting you now.").dial(
 
 ## SDK Architecture
 
-The SDK auto-generates verb methods from `specs.json` (from `@jambonz/verb-specifications`). When the spec changes, the SDK automatically picks up new parameters:
+The SDK auto-generates verb methods from JSON Schema files (from `@jambonz/schema`). When the schema changes, the SDK automatically picks up new parameters:
 
-1. `specs.json` — bundled verb/component specifications (synced from upstream)
+1. `schema/verbs/*.schema.json` — bundled verb schemas (synced from upstream)
 2. `verb_registry.py` — maps spec entries to Python methods + synonyms
 3. `verb_builder.py` — generates methods at import time from specs + registry
 4. `WebhookResponse` and `Session` both extend `VerbBuilder`

CLAUDE.md

@@ -23,7 +23,7 @@ src/jambonz_sdk/
 │   ├── verbs.py         # All 26+ verb TypedDicts
 │   ├── rest.py          # REST API request/response types
 │   └── session.py       # Call session & WebSocket message types
-├── verb_builder.py      # VerbBuilder — methods auto-generated from specs.json
+├── verb_builder.py      # VerbBuilder — methods auto-generated from JSON Schema
 ├── verb_registry.py     # Verb definitions: maps spec entries → Python methods
 ├── webhook/
 │   ├── __init__.py
@@ -48,12 +48,12 @@ src/jambonz_sdk/
 - **Transport-agnostic verb building**: Same verb methods on both `WebhookResponse` and `Session`
 - **Fluent/chainable API**: All verb methods return `self` for method chaining
 - **TypedDict for verb schemas**: Type-safe verb construction matching JSON schemas exactly
-- **Auto-generated verb methods**: VerbBuilder methods are generated at import time from `specs.json` + `verb_registry.py` — when the spec changes, the SDK automatically picks up new parameters
+- **Auto-generated verb methods**: VerbBuilder methods are generated at import time from JSON Schema files (`@jambonz/schema`) + `verb_registry.py` — when the schema changes, the SDK automatically picks up new parameters
 - **aiohttp for both HTTP and WebSocket**: Single dependency for REST client and WS transport
 
 ## Verb System
 
-The SDK supports all 26+ jambonz verbs. Verb methods on VerbBuilder are **auto-generated** from the shared `specs.json` (in `/Users/xhoaluu/jambonz/verb-specifications/specs.json`).
+The SDK supports all 26+ jambonz verbs. Verb methods on VerbBuilder are **auto-generated** from JSON Schema files bundled from [`@jambonz/schema`](https://github.com/jambonz/schema).
 
 ### How verb generation works
 
@@ -65,7 +65,7 @@ The SDK supports all 26+ jambonz verbs. Verb methods on VerbBuilder are **auto-g
 ### Verb List
 
 Audio/Speech: `say`, `play`, `gather`
-AI/S2S: `openai_s2s`, `google_s2s`, `deepgram_s2s`, `elevenlabs_s2s`, `ultravox_s2s`, `s2s`, `llm`, `dialogflow`, `pipeline`
+AI/S2S: `openai_s2s`, `google_s2s`, `deepgram_s2s`, `elevenlabs_s2s`, `ultravox_s2s`, `s2s`, `llm`, `dialogflow`, `agent`
 Call Control: `dial`, `conference`, `enqueue`, `dequeue`, `hangup`, `redirect`, `pause`
 Audio Streaming: `listen`, `stream`, `transcribe`
 SIP: `sip_decline`, `sip_request`, `sip_refer`
@@ -105,7 +105,7 @@ Source: https://github.com/jambonz/schema
 
 `AGENTS.md` is the comprehensive developer guide for AI agents working with this SDK.
 It covers: verb system, webhook/WebSocket patterns, REST API, env vars, mid-call control,
-TTS streaming, pipeline updates, audio streaming, and common application patterns.
+TTS streaming, agent updates, audio streaming, and common application patterns.
 AI coding agents should read AGENTS.md before generating jambonz Python application code.
 
 ### MCP Server
@@ -171,11 +171,11 @@ pytest                      # All 279 tests
 ### Unit tests (`tests/unit/`)
 - `test_verb_builder.py` — Parametrized across all 31 verb defs: method existence, correct verb name, all spec properties pass through
 - `test_webhook.py` — Webhook contract, HMAC-SHA256 signature protocol, env vars OPTIONS format
-- `test_session.py` — WebSocket protocol messages: ack, command, tts:tokens, llm:tool-output, pipeline:update
+- `test_session.py` — WebSocket protocol messages: ack, command, tts:tokens, llm:tool-output, agent:update
 - `test_ws_client.py` — Message routing: session:new, verb:hook dispatch, auto-reply, binary/JSON robustness
 - `test_rest_client.py` — REST API contract: URL construction, HTTP methods, request bodies
 - `test_audio_stream.py` — Audio protocol: raw PCM, playAudio JSON, marks, control commands
 
 ### Integration tests (`tests/integration/`)
 - `test_webhook.py` — Real aiohttp server with IVR menu, actionHook routing, env vars discovery
-- `test_websocket.py` — Real WebSocket connections: full protocol compliance, multi-step conversations, inject commands, TTS streaming, pipeline updates
+- `test_websocket.py` — Real WebSocket connections: full protocol compliance, multi-step conversations, inject commands, TTS streaming, agent updates

README.md

@@ -72,7 +72,7 @@ async with JambonzClient(
 
 ### Spec-driven verb generation
 
-The SDK does **not** hardcode verb method signatures. Instead, verb methods (`.say()`, `.gather()`, `.dial()`, `.pipeline()`, etc.) are **auto-generated at import time** from [JSON Schema](https://github.com/jambonz/schema) files — the same schemas used by the Node.js SDK and the jambonz server.
+The SDK does **not** hardcode verb method signatures. Instead, verb methods (`.say()`, `.gather()`, `.dial()`, `.agent()`, etc.) are **auto-generated at import time** from [JSON Schema](https://github.com/jambonz/schema) files — the same schemas used by the Node.js SDK and the jambonz server.
 
 **What this means:**
 
@@ -98,15 +98,15 @@ VerbDef("new_verb", "new_verb", doc="Description.")
 
 ## Features
 
-- **All 31 jambonz verbs**: say, play, gather, dial, conference, enqueue/dequeue, hangup, pause, redirect, config, tag, dtmf, dub, message, alert, answer, leave, listen/stream, transcribe, openai_s2s, google_s2s, deepgram_s2s, elevenlabs_s2s, ultravox_s2s, s2s, llm, dialogflow, pipeline, sip_decline, sip_request, sip_refer
+- **All 31 jambonz verbs**: say, play, gather, dial, conference, enqueue/dequeue, hangup, pause, redirect, config, tag, dtmf, dub, message, alert, answer, leave, listen/stream, transcribe, openai_s2s, google_s2s, deepgram_s2s, elevenlabs_s2s, ultravox_s2s, s2s, llm, dialogflow, agent, sip_decline, sip_request, sip_refer
 - **Fluent chainable API**: `.say(...).gather(...).hangup()`
 - **Webhook transport**: `WebhookResponse` for HTTP apps (works with aiohttp, FastAPI, Flask, etc.)
 - **WebSocket transport**: `create_endpoint` with `Session`, event handling, `send()`/`reply()`
 - **REST client**: `JambonzClient` with calls, conferences, queues, mid-call control
 - **Audio streaming**: Bidirectional audio via `AudioStream`
 - **Mid-call control**: inject commands (mute, whisper, record, DTMF, tag)
 - **TTS token streaming**: `send_tts_tokens()` / `flush_tts_tokens()`
-- **Pipeline updates**: `update_pipeline()` for mid-conversation LLM changes
+- **Agent updates**: `update_agent()` for mid-conversation LLM changes
 - **Signature verification**: HMAC-SHA256 webhook signature validation
 - **Env vars**: Portal discovery via OPTIONS + runtime reading
 
@@ -119,7 +119,7 @@ See the [`examples/`](examples/) directory:
 | hello-world | [webhook](examples/hello-world/webhook_app.py) | [websocket](examples/hello-world/websocket_app.py) | Minimal greeting |
 | echo | [webhook](examples/echo/webhook_app.py) | [websocket](examples/echo/websocket_app.py) | Speech echo with gather |
 | ivr-menu | [webhook](examples/ivr-menu/webhook_app.py) | — | IVR menu with speech + DTMF |
-| voice-agent | [webhook](examples/voice-agent/webhook_app.py) | [websocket](examples/voice-agent/websocket_app.py) | LLM pipeline with tool calls |
+| voice-agent | [webhook](examples/voice-agent/webhook_app.py) | [websocket](examples/voice-agent/websocket_app.py) | LLM agent with tool calls |
 | dial | [webhook](examples/dial/webhook_app.py) | — | Outbound dial with fallback |
 | listen-record | [webhook](examples/listen-record/webhook_app.py) | [websocket](examples/listen-record/websocket_app.py) | Audio recording |
 

examples/voice-agent/websocket_app.py

@@ -1,7 +1,7 @@
 """Voice Agent - WebSocket example.
 
-LLM-powered voice agent using the pipeline verb with tool calling.
-Demonstrates pipeline configuration, eventHook handling, and toolHook handling.
+LLM-powered voice agent using the agent verb with tool calling.
+Demonstrates agent configuration, eventHook handling, and toolHook handling.
 
 Usage:
     python websocket_app.py
@@ -40,7 +40,7 @@ async def handle_session(session):
 
         print(f"New call: {session.call_sid} from {session.from_}")
 
-        # Handle pipeline events
+        # Handle agent events
         async def on_event(evt):
             event_type = evt.get("type", "")
             if event_type == "turn_end":
@@ -51,7 +51,7 @@ async def on_event(evt):
                 )
             await session.reply()
 
-        session.on("/pipeline-event", on_event)
+        session.on("/agent-event", on_event)
 
         # Handle tool calls
         async def on_tool(evt):
@@ -71,16 +71,16 @@ async def on_tool(evt):
 
         session.on("/tool-call", on_tool)
 
-        # Handle pipeline completion
+        # Handle agent completion
         async def on_complete(evt):
-            print(f"Pipeline complete: {evt.get('completion_reason', 'unknown')}")
+            print(f"Agent complete: {evt.get('completion_reason', 'unknown')}")
             session.hangup()
             await session.reply()
 
-        session.on("/pipeline-complete", on_complete)
+        session.on("/agent-complete", on_complete)
 
-        # Start the pipeline
-        session.pipeline(
+        # Start the agent
+        session.agent(
             stt={
                 "vendor": "deepgram",
                 "language": "en-US",
@@ -133,9 +133,9 @@ async def on_complete(evt):
             turnDetection="krisp",
             earlyGeneration=True,
             bargeIn={"enable": True, "minSpeechDuration": 0.3},
-            eventHook="/pipeline-event",
+            eventHook="/agent-event",
             toolHook="/tool-call",
-            actionHook="/pipeline-complete",
+            actionHook="/agent-complete",
         )
         await session.send()
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "jambonz-python-sdk"
-version = "0.2.0"
+version = "0.3.0"
 description = "Python SDK for jambonz CPaaS platform"
 readme = "README.md"
 requires-python = ">=3.10"

scripts/generate_stubs.py

@@ -1,9 +1,9 @@
 #!/usr/bin/env python3
-"""Generate verb_builder.pyi stub file from specs.json + verb_registry.
+"""Generate verb_builder.pyi stub file from JSON Schema + verb_registry.
 
 This creates a .pyi type stub that IDEs (VS Code Pylance, PyCharm, mypy)
 read for static type checking and autocomplete. Run this after syncing
-specs.json or updating verb_registry.py.
+the schema or updating verb_registry.py.
 
 Usage:
     python scripts/generate_stubs.py
@@ -19,10 +19,10 @@
 
 from jambonz_sdk.verb_registry import VERB_DEFS
 
-SPECS_PATH = SRC_DIR / "jambonz_sdk" / "specs.json"
+SCHEMA_DIR = SRC_DIR / "jambonz_sdk" / "schema" / "verbs"
 STUB_PATH = SRC_DIR / "jambonz_sdk" / "verb_builder.pyi"
 
-# Maps specs.json type strings to Python type annotation strings for .pyi
+# Maps JSON Schema type strings to Python type annotation strings for .pyi
 TYPE_MAP = {
     "string": "str",
     "number": "int | float",
@@ -33,7 +33,7 @@
 
 
 def resolve_type(spec_type) -> str:
-    """Convert a specs.json type descriptor to a .pyi type string."""
+    """Convert a JSON Schema type descriptor to a .pyi type string."""
     if isinstance(spec_type, str):
         if spec_type.startswith("#"):
             return "dict[str, Any]"
@@ -61,9 +61,37 @@ def resolve_type(spec_type) -> str:
     return "Any"
 
 
+def _load_schemas() -> dict:
+    """Load verb JSON Schemas from the bundled schema directory."""
+    schemas: dict = {}
+    for schema_file in sorted(SCHEMA_DIR.glob("*.schema.json")):
+        with schema_file.open() as f:
+            schema = json.load(f)
+        schema_id = schema.get("$id", "")
+        if schema_id:
+            spec_name = schema_id.rsplit("/", 1)[-1]
+        else:
+            spec_name = schema_file.stem.replace(".schema", "")
+        properties = {}
+        for prop_name, prop_def in schema.get("properties", {}).items():
+            if prop_name == "verb":
+                continue
+            properties[prop_name] = prop_def
+        for entry in schema.get("allOf", []):
+            if "properties" in entry:
+                for prop_name, prop_def in entry["properties"].items():
+                    if prop_name == "verb":
+                        continue
+                    properties[prop_name] = prop_def
+        schemas[spec_name] = {
+            "properties": properties,
+            "required": schema.get("required", []),
+        }
+    return schemas
+
+
 def generate() -> str:
-    with SPECS_PATH.open() as f:
-        specs = json.load(f)
+    specs = _load_schemas()
 
     lines = [
         '"""Auto-generated type stubs for VerbBuilder.',
@@ -75,7 +103,6 @@ def generate() -> str:
         "",
         "from jambonz_sdk.types.verbs import AnyVerb",
         "",
-        "",
         "class VerbBuilder:",
         "    _verbs: list[AnyVerb]",
         "",

scripts/sync_schema.py

@@ -22,7 +22,7 @@
 from pathlib import Path
 
 # ── Pin the schema version here ──────────────────────────────────────
-SCHEMA_VERSION = "v0.1.1"
+SCHEMA_VERSION = "v0.2.1"
 # ────────────────────────────────────────────────────────────────────
 
 DEST = Path(__file__).resolve().parent.parent / "src" / "jambonz_sdk" / "schema"

src/jambonz_sdk/client/api.py

@@ -100,16 +100,16 @@ async def mute(self, call_sid: str, status: str) -> dict[str, Any]:
         """
         return await self.update(call_sid, {"mute_status": status})
 
-    async def update_pipeline(
+    async def update_agent(
         self, call_sid: str, data: dict[str, Any]
     ) -> dict[str, Any]:
-        """Send a mid-conversation pipeline update.
+        """Send a mid-conversation agent update.
 
         Args:
             call_sid: The call to update.
-            data: Pipeline update payload.
+            data: Agent update payload.
         """
-        return await self.update(call_sid, {"pipeline_update": data})
+        return await self.update(call_sid, {"agent_update": data})
 
     async def noise_isolation(
         self, call_sid: str, status: str, opts: dict[str, Any] | None = None

src/jambonz_sdk/schema/callbacks/pipeline-turn.schema.json → src/jambonz_sdk/schema/callbacks/agent-turn.schema.json

@@ -1,8 +1,8 @@
 {
   "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "$id": "https://jambonz.org/schema/callbacks/pipeline-turn",
-  "title": "Pipeline EventHook Events",
-  "description": "Events sent to the pipeline verb's eventHook during a conversation. These are sent as 'pipeline:event' messages over the WebSocket connection.",
+  "$id": "https://jambonz.org/schema/callbacks/agent-turn",
+  "title": "Agent EventHook Events",
+  "description": "Events sent to the agent verb's eventHook during a conversation. These are sent as 'agent:event' messages over the WebSocket connection.",
   "type": "object",
   "oneOf": [
     {
@@ -84,7 +84,7 @@
     {
       "properties": {
         "type": {
-          "const": "agent_response",
+          "const": "llm_response",
           "description": "Sent when the LLM has finished generating its response for the current turn. Contains the complete response text."
         },
         "response": {