@moda-ai/cli 1.10.2 → 1.12.0

package/skills/integration/integration-python-generic/EXAMPLE.md

@@ -0,0 +1,46 @@
+# Golden reference: generic Python app + Moda
+
+The complete finished integration for a small Python chat script whose LLM
+calls go through the `openai` client (automatically instrumented). Match
+this structure; adapt names to the target repository.
+
+## app/main.py
+
+```python
+import os
+
+import moda
+from moda import set_conversation_id, set_user_id
+from openai import OpenAI
+
+# Initialize Moda once, before any LLM client is created.
+moda.init(os.environ["MODA_API_KEY"])
+
+client = OpenAI()
+
+
+def answer(session_id: str, user_id: str, message: str) -> str:
+    # Scoped context: safe under concurrent callers.
+    with set_conversation_id(f"session_{session_id}"):
+        with set_user_id(user_id):
+            response = client.chat.completions.create(
+                model="gpt-4o",
+                messages=[{"role": "user", "content": message}],
+            )
+    return response.choices[0].message.content
+
+
+print(answer("demo-session", "user-123", "Hello!"))
+
+# Flush pending telemetry before exit.
+moda.flush()
+```
+
+## What changed, and what did not
+
+- `moda-ai` added to the project dependencies.
+- One `moda.init` call added at startup, before any LLM client is created.
+- Call sites through instrumented clients are unchanged; only context
+  scoping was added around them.
+- No `.env` file was created or edited; `MODA_API_KEY` is read from the
+  runtime environment.

package/skills/integration/integration-python-generic/SKILL.md

@@ -0,0 +1,146 @@
+---
+id: integration-python-generic
+name: "Integrate the Moda SDK (Python, any provider)"
+version: 1.0.0
+category: integration
+targets:
+  language: python
+  frameworks: []
+  providers: []
+sdk_package: moda-ai
+summary: "General Moda SDK integration for a Python app when no specific framework or provider skill matches."
+---
+
+# Integrate the Moda SDK (Python, any provider)
+
+This is the general Python integration path, used when no framework- or
+provider-specific skill matches. Follow the steps below in order. Each step
+has a stop condition; when you hit one, stop and report instead of guessing.
+`EXAMPLE.md` in this directory shows a complete finished integration; match
+its structure. For API details beyond this skill, read
+https://docs.moda.dev/ingestion/moda-sdk.
+
+## Safety rules (apply to every step)
+
+- Only touch the repository you were launched in.
+- Never write the `MODA_API_KEY` value (or any secret) into source files, and
+  never create or edit `.env` files. The key belongs in the runtime
+  environment or a secret manager.
+- Make the smallest correct change. Match the existing code style and imports.
+- Do not commit unless the user asks.
+- If a step is ambiguous in this repository, stop and report rather than
+  guessing.
+
+## Step 1: Find the real LLM call sites and pick the right surface
+
+Search the repository for LLM calls and classify them:
+
+- Calls through an instrumented provider or framework: automatically
+  captured once Moda is initialized (Step 3); no per-call changes. The
+  Python SDK's `Instruments` set covers providers including OpenAI,
+  Anthropic, Cohere, Mistral, Groq, Ollama, Bedrock, VertexAI, and
+  frameworks including LangChain, LlamaIndex, Haystack, and CrewAI (full
+  table: https://docs.moda.dev/ingestion/moda-sdk). OpenAI-compatible
+  gateways (for example OpenRouter used through the `openai` client with a
+  custom `base_url`) count as OpenAI here.
+- Calls through LangChain or the Claude Agent SDK: stop — use the dedicated
+  skill instead; never instrument the underlying provider client when a
+  wrapper framework is in use.
+- Direct HTTP calls to a provider outside the instruments table: the Python
+  SDK does not capture these automatically. Note them in your report and
+  point the user at https://docs.moda.dev/ingestion/direct-api; do not
+  build a custom ingestion path yourself.
+
+If the repository has no LLM call sites at all, stop and report.
+
+## Step 2: Install the SDK
+
+Install `moda-ai` with the package manager this repository already uses
+(`uv.lock` means uv, `poetry.lock` means poetry, otherwise pip):
+
+```bash
+pip install moda-ai
+```
+
+With uv:
+
+```bash
+uv add moda-ai
+```
+
+With poetry:
+
+```bash
+poetry add moda-ai
+```
+
+If `moda-ai` is already a dependency, skip the install.
+
+## Step 3: Initialize Moda once at startup
+
+`moda.init` is synchronous and must run once, at startup, before any LLM
+client is created. If a `moda.init` call already exists, verify it runs at
+startup and skip this step; never add a second call.
+
+```python
+import os
+
+import moda
+
+moda.init(os.environ["MODA_API_KEY"])
+```
+
+## Step 4: Set conversation and user context
+
+Set a stable conversation id before each LLM call so multi-turn
+conversations group together, and a user id when one user can be identified.
+If one user reaches the same agent through several channels, use the same
+user id across channels.
+
+For sequential code, use the module-level properties:
+
+```python
+moda.conversation_id = f"session_{session_id}"
+moda.user_id = user_id
+```
+
+For concurrent request handlers (any web server), use the scoped context
+managers so overlapping requests do not leak context into each other:
+
+```python
+from moda import set_conversation_id, set_user_id
+
+with set_conversation_id(f"session_{session_id}"):
+    with set_user_id(user_id):
+        response = make_llm_call(user_message)
+```
+
+Reuse the session/thread identifier the app already has. If no session
+identifier exists anywhere, the SDK falls back to hashing the first user
+message plus the system prompt — acceptable only for simple chatbots that
+pass full message history on every call. If this repo is an agent framework
+or does not pass full history, an explicit conversation id is required; stop
+and report if you cannot find a session identifier to use.
+
+## Step 5: Flush before exit
+
+Flush pending telemetry before the process exits (short-lived scripts) or on
+shutdown (servers):
+
+```python
+moda.flush()
+```
+
+## Step 6: Validate
+
+Report the packages installed, files edited, and any step a human still needs
+to do (for example setting `MODA_API_KEY` in the deploy environment). Then
+verify telemetry reaches Moda:
+
+```bash
+moda doctor --online --json
+```
+
+The report should show recent traces. If it does not, confirm `MODA_API_KEY`
+is available in the runtime environment and that `moda.init()` runs before
+the first LLM client is created.

package/skills/integration/integration-python-langchain/EXAMPLE.md

@@ -0,0 +1,63 @@
+# Golden reference: Python LangChain agent + Moda
+
+The complete finished integration for a minimal LangChain app. Match this
+structure; adapt names to the target repository. Note what is absent: no
+LangChain callbacks or handler classes for Moda — `moda.init()` enables all
+instruments (including LangChain) by default, and the only per-run change is
+the explicit conversation id wrapper.
+
+## app/agent.py
+
+```python
+import os
+
+import moda
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_openai import ChatOpenAI
+from moda import set_conversation_id, set_user_id
+
+# Initialize Moda once, before LangChain models are constructed. moda.init
+# enables all instruments (including LangChain) by default.
+moda.init(os.environ["MODA_API_KEY"])
+
+model = ChatOpenAI(model="gpt-4o")
+prompt = ChatPromptTemplate.from_messages([
+    ("system", "You are a helpful assistant."),
+    ("human", "{input}"),
+])
+chain = prompt | model
+
+
+def run_agent(session_id: str, user_id: str, user_query: str):
+    # Explicit conversation id around the whole run: every internal LLM call
+    # made during this invocation shares it. Automatic detection does NOT
+    # work for agent frameworks.
+    with set_conversation_id(f"agent_session_{session_id}"):
+        with set_user_id(user_id):
+            return chain.invoke({"input": user_query})
+```
+
+## app/main.py
+
+```python
+import moda
+
+from app.agent import run_agent
+
+result = run_agent("demo-session", "user-123", "What is Python?")
+print(result.content)
+
+# Flush pending telemetry before exit.
+moda.flush()
+```
+
+## What changed, and what did not
+
+- `moda-ai` added to the project dependencies.
+- One `moda.init` call added at startup, before any LangChain model is
+  constructed.
+- Each agent/chain entry point is wrapped in the scoped context managers
+  with an explicit conversation id; the LangChain code itself is unchanged.
+- No LangChain callbacks or handler classes were added for Moda.
+- No `.env` file was created or edited; `MODA_API_KEY` is read from the
+  runtime environment.

package/skills/integration/integration-python-langchain/SKILL.md

@@ -0,0 +1,141 @@
+---
+id: integration-python-langchain
+name: "Integrate Moda with LangChain / LangGraph (Python)"
+version: 1.0.0
+category: integration
+targets:
+  language: python
+  frameworks:
+    - langchain
+  providers: []
+sdk_package: moda-ai
+summary: "Instrument a Python LangChain or LangGraph app with Moda: init before the framework, explicit conversation ids around every run."
+---
+
+# Integrate Moda with LangChain / LangGraph (Python)
+
+Follow the steps below in order. Each step has a stop condition; when you hit
+one, stop and report instead of guessing. `EXAMPLE.md` in this directory shows
+the complete finished integration; match its structure. For API details beyond
+this skill, read https://docs.moda.dev/ingestion/moda-sdk.
+
+## Safety rules (apply to every step)
+
+- Only touch the repository you were launched in.
+- Never write the `MODA_API_KEY` value (or any secret) into source files, and
+  never create or edit `.env` files. The key belongs in the runtime
+  environment or a secret manager.
+- Make the smallest correct change. Match the existing code style and imports.
+- Do not commit unless the user asks.
+- If a step is ambiguous in this repository, stop and report rather than
+  guessing.
+
+## What is different about agent frameworks
+
+Two constraints drive every step below:
+
+1. Instrument at the Moda SDK level, not inside LangChain. Moda's Python SDK
+   ships a LangChain instrument (`Instruments.LANGCHAIN`), and `moda.init()`
+   enables all instruments by default (`instruments=None`). Do not add
+   LangChain callbacks or handler classes for Moda — no such Moda API
+   exists.
+2. Explicit conversation ids are mandatory. Moda's automatic conversation
+   detection hashes the first user message, and agent frameworks construct
+   different messages on every internal iteration, so without an explicit id
+   every iteration lands in a different conversation.
+
+## Step 1: Find the agent/chain entry points
+
+Find where the application invokes LangChain or LangGraph: `.invoke(...)`,
+`.stream(...)`, agent executor runs, or graph runs. These entry points — not
+the internal LLM calls — are where conversation context gets set.
+
+If you cannot find any LangChain/LangGraph entry point, stop and report.
+
+## Step 2: Install the SDK
+
+Install `moda-ai` with the package manager this repository already uses
+(`uv.lock` means uv, `poetry.lock` means poetry, otherwise pip):
+
+```bash
+pip install moda-ai
+```
+
+With uv:
+
+```bash
+uv add moda-ai
+```
+
+With poetry:
+
+```bash
+poetry add moda-ai
+```
+
+If `moda-ai` is already a dependency, skip the install.
+
+## Step 3: Initialize Moda once at startup
+
+`moda.init` is synchronous and must run before LangChain models, chains, or
+graphs are constructed. Initialize at the top of the application entry
+point. If a `moda.init` call already exists, verify it runs at startup and
+skip this step; never add a second call.
+
+```python
+import os
+
+import moda
+
+moda.init(os.environ["MODA_API_KEY"])
+```
+
+## Step 4: Wrap every agent/chain run in an explicit conversation id
+
+Wrap each entry point found in Step 1 so all internal LLM calls made during
+that run share one conversation id. Use the scoped context managers so
+concurrent runs do not leak context into each other:
+
+```python
+from moda import set_conversation_id, set_user_id
+
+with set_conversation_id(f"agent_session_{session_id}"):
+    with set_user_id(user_id):
+        result = agent.invoke({"input": user_query})
+```
+
+For strictly sequential code, the module-level property form is equivalent:
+
+```python
+moda.conversation_id = f"agent_session_{session_id}"
+result = agent.invoke({"input": user_query})
+moda.conversation_id = None
+```
+
+Reuse the session/thread identifier the app already has (LangGraph
+`thread_id`, a chat session id, a ticket id). If no session identifier
+exists anywhere, stop and report; do not invent one.
+
+## Step 5: Flush before exit
+
+Flush pending telemetry before the process exits (short-lived scripts) or on
+shutdown (servers):
+
+```python
+moda.flush()
+```
+
+## Step 6: Validate
+
+Report the packages installed, files edited, and any step a human still needs
+to do (for example setting `MODA_API_KEY` in the deploy environment). Then
+verify telemetry reaches Moda:
+
+```bash
+moda doctor --online --json
+```
+
+The report should show recent traces, and multi-step agent runs should share
+one conversation id. If each internal LLM call shows up as its own
+conversation, the explicit conversation id from Step 4 is not wrapping the
+run — fix that before finishing.

package/skills/integration/integration-python-openai/EXAMPLE.md

@@ -0,0 +1,65 @@
+# Golden reference: FastAPI + OpenAI SDK + Moda
+
+The complete finished integration for a minimal FastAPI chat service that
+calls the OpenAI SDK directly. Match this structure; adapt names to the
+target repository.
+
+## app/main.py
+
+```python
+import os
+from contextlib import asynccontextmanager
+
+import moda
+from fastapi import FastAPI
+from moda import set_conversation_id, set_user_id
+from openai import OpenAI
+from pydantic import BaseModel
+
+# Initialize Moda before creating any LLM client so instrumentation is
+# active. After this, every chat.completions.create call is captured
+# automatically.
+moda.init(os.environ["MODA_API_KEY"])
+
+client = OpenAI()
+
+
+@asynccontextmanager
+async def lifespan(app: FastAPI):
+    yield
+    # Flush pending telemetry on shutdown.
+    moda.flush()
+
+
+app = FastAPI(lifespan=lifespan)
+
+
+class ChatRequest(BaseModel):
+    session_id: str
+    user_id: str
+    message: str
+
+
+@app.post("/chat")
+def chat(request: ChatRequest):
+    # Scoped context: safe under concurrent requests.
+    with set_conversation_id(f"session_{request.session_id}"):
+        with set_user_id(request.user_id):
+            response = client.chat.completions.create(
+                model="gpt-4o",
+                messages=[{"role": "user", "content": request.message}],
+            )
+    return {"reply": response.choices[0].message.content}
+```
+
+## What changed, and what did not
+
+- `moda-ai` added to the project dependencies.
+- One `moda.init` call added at module import time, before `OpenAI()` is
+  constructed.
+- The handler wraps its existing OpenAI call in the scoped context managers;
+  the `chat.completions.create` call itself is unchanged — instrumentation
+  is automatic.
+- Shutdown flushes telemetry via the FastAPI lifespan hook.
+- No `.env` file was created or edited; `MODA_API_KEY` is read from the
+  runtime environment.

package/skills/integration/integration-python-openai/SKILL.md

@@ -0,0 +1,151 @@
+---
+id: integration-python-openai
+name: "Integrate Moda with the OpenAI SDK (Python)"
+version: 1.0.0
+category: integration
+targets:
+  language: python
+  frameworks: []
+  providers:
+    - openai
+sdk_package: moda-ai
+summary: "Auto-instrument openai client calls with Moda in a Python app; no per-call changes required."
+---
+
+# Integrate Moda with the OpenAI SDK (Python)
+
+Follow the steps below in order. Each step has a stop condition; when you hit
+one, stop and report instead of guessing. `EXAMPLE.md` in this directory shows
+the complete finished integration; match its structure. For API details beyond
+this skill, read https://docs.moda.dev/ingestion/providers/openai and
+https://docs.moda.dev/ingestion/moda-sdk.
+
+## Safety rules (apply to every step)
+
+- Only touch the repository you were launched in.
+- Never write the `MODA_API_KEY` value (or any secret) into source files, and
+  never create or edit `.env` files. The key belongs in the runtime
+  environment or a secret manager.
+- Make the smallest correct change. Match the existing code style and imports.
+- Do not commit unless the user asks.
+- If a step is ambiguous in this repository, stop and report rather than
+  guessing.
+
+## Step 1: Confirm the raw OpenAI client is the integration surface
+
+Find the real LLM call sites: `client.chat.completions.create(...)` on a
+client constructed with `OpenAI()` from the `openai` package.
+
+- If the OpenAI calls go through a wrapper framework instead (LangChain,
+  CrewAI, the Claude Agent SDK, or similar), stop: instrument the wrapper
+  layer with its dedicated skill, never the underlying provider client.
+- If the repo wraps its OpenAI client in a helper module (for example
+  `app/llm.py` exporting a shared client), plan to instrument at that one
+  place; do not edit every call site.
+- If there are no OpenAI call sites at all, stop and report.
+
+## Step 2: Install the SDK
+
+Install `moda-ai` with the package manager this repository already uses
+(`uv.lock` means uv, `poetry.lock` means poetry, otherwise pip):
+
+```bash
+pip install moda-ai
+```
+
+With uv:
+
+```bash
+uv add moda-ai
+```
+
+With poetry:
+
+```bash
+poetry add moda-ai
+```
+
+If `moda-ai` is already a dependency, skip the install. `moda-ai` includes
+the OpenAI and Anthropic instrumentation packages; nothing else is needed
+for this integration.
+
+## Step 3: Initialize Moda once at startup, before the client is created
+
+Instrumentation is automatic: once `moda.init` has run, every
+`chat.completions.create` call (including streaming and tool use) is
+captured with no per-call changes.
+
+`moda.init` is synchronous and must run before the OpenAI client is
+constructed. If a `moda.init` call already exists, verify it runs at startup
+before client creation and skip this step; never add a second call.
+
+```python
+import os
+
+import moda
+from openai import OpenAI
+
+moda.init(os.environ["MODA_API_KEY"])
+
+client = OpenAI()
+```
+
+For web frameworks, initialize at import time of the application module (as
+above) so the client-creation ordering holds; requirement: `moda.init` runs
+before any `OpenAI()` construction.
+
+## Step 4: Set conversation and user context
+
+Set a stable conversation id before each LLM call so multi-turn
+conversations group together, and a user id when one user can be identified.
+
+For sequential code, use the module-level properties:
+
+```python
+moda.conversation_id = f"session_{session_id}"
+moda.user_id = user_id
+```
+
+For concurrent request handlers (any web server), use the scoped context
+managers so overlapping requests do not leak context into each other:
+
+```python
+from moda import set_conversation_id, set_user_id
+
+with set_conversation_id(f"session_{session_id}"):
+    with set_user_id(user_id):
+        response = client.chat.completions.create(
+            model="gpt-4o",
+            messages=[{"role": "user", "content": user_message}],
+        )
+```
+
+Reuse the session/thread identifier the app already has. If no session
+identifier exists anywhere, the SDK falls back to hashing the first user
+message plus the system prompt — acceptable only for simple chatbots that
+pass full message history on every call. If this repo is an agent framework
+or does not pass full history, an explicit conversation id is required; stop
+and report if you cannot find a session identifier to use.
+
+## Step 5: Flush before exit
+
+Flush pending telemetry before the process exits (short-lived scripts) or on
+shutdown (servers):
+
+```python
+moda.flush()
+```
+
+## Step 6: Validate
+
+Report the packages installed, files edited, and any step a human still needs
+to do (for example setting `MODA_API_KEY` in the deploy environment). Then
+verify telemetry reaches Moda:
+
+```bash
+moda doctor --online --json
+```
+
+The report should show recent traces. If it does not, confirm `MODA_API_KEY`
+is available in the runtime environment and that `moda.init()` runs before
+the OpenAI client is created.