PLG-Motion | AutoGraph Flow Tutorial - #1061
Conversation
|
Deploy Preview Available Via |
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
📝 WalkthroughWalkthroughAdded two Contextual Data Platform tutorials and a complete AutoGraph demonstration notebook. The content covers local Kubernetes installation, platform configuration, service deployment, document ingestion, GraphRAG orchestration, querying, validation, and cleanup. ChangesContextual Data Platform tutorials
Estimated code review effort: 3 (Moderate) | ~25 minutes Possibly related PRs
Suggested labels: Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
🧹 Nitpick comments (1)
site/content/contextual-data-platform/tutorials/evaluate-locally.md (1)
340-340: 🩺 Stability & Availability | 🔵 Trivial | ⚡ Quick winPin the MinIO images used by the tutorial.
minio/minio:latestand the untaggedminio/mcimage can change between runs, causing a previously validated tutorial to fail or behave differently. Pin both to tested, compatible versions or immutable digests.Also applies to: 381-381
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@site/content/contextual-data-platform/tutorials/evaluate-locally.md` at line 340, Pin the MinIO images in the tutorial’s container configuration, including the `minio/minio` entry and the `minio/mc` entry at the referenced second location, to tested compatible version tags or immutable digests. Replace `minio/minio:latest` and the untagged `minio/mc` reference while preserving the existing service configuration.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@site/content/contextual-data-platform/tutorials/evaluate-locally.md`:
- Around line 129-132: Add the text language identifier to the output-only
Markdown fences around the kubectl output blocks, including the sections
corresponding to the listed ranges. Keep the existing output content unchanged
and ensure every affected fenced block has an explicit language tag to satisfy
markdownlint MD040.
In `@site/content/contextual-data-platform/tutorials/notebook-tutorial.md`:
- Around line 101-116: The tutorial configuration in
site/content/contextual-data-platform/tutorials/notebook-tutorial.md lines
101-116 must expose the provider, model, and API URL settings, or explicitly
state that OpenAI is required. Update site/static/notebooks/Autograph_DEMO.ipynb
lines 295-306 to read those provider settings from os.environ instead of
hard-coding them, keeping the documented variable names consistent across both
files.
- Around line 432-449: The notebook tutorial must wait for the asynchronous
import orchestration to finish before proceeding to Retriever deployment; update
the workflow around the orchestration ID and Step 12 in
site/content/contextual-data-platform/tutorials/notebook-tutorial.md (lines
432-449) to require a completed status. In
site/static/notebooks/Autograph_DEMO.ipynb (lines 567-720), add explicit polling
that continues on in-progress status, handles completed and failed outcomes, and
only stops AutoGraph or proceeds after completion.
- Around line 352-356: Update the Step 10 warning in the notebook tutorial to
direct users to inspect the build-status response’s message and error_code
fields when status is failed; remove the reference to the unsupported error
field while preserving the existing guidance about builds still in progress.
- Around line 92-97: Add requests to the notebook setup installation in
site/static/notebooks/Autograph_DEMO.ipynb at lines 65-85 before its import, and
update the prerequisite statement in
site/content/contextual-data-platform/tutorials/notebook-tutorial.md at lines
92-97 to accurately mention requests or otherwise remove the claim that only
python-dotenv is required.
In `@site/static/notebooks/Autograph_DEMO.ipynb`:
- Around line 895-902: Update the Deep Search request body in the notebook cell
to explicitly set use_llm_planner to true alongside query_type: 2, ensuring it
follows the Deep Search contract without relying on a default.
- Around line 123-165: Update authenticate() and send_request() to pass a finite
timeout to their requests.post/request calls, and apply the same timeout to
every requests.post and requests.get call in the optional metadata cell. Reuse
the notebook’s existing timeout configuration if available; otherwise define one
shared finite value and use it consistently.
---
Nitpick comments:
In `@site/content/contextual-data-platform/tutorials/evaluate-locally.md`:
- Line 340: Pin the MinIO images in the tutorial’s container configuration,
including the `minio/minio` entry and the `minio/mc` entry at the referenced
second location, to tested compatible version tags or immutable digests. Replace
`minio/minio:latest` and the untagged `minio/mc` reference while preserving the
existing service configuration.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 90889f44-0703-4e8a-8dd4-e79e12f6a8cd
⛔ Files ignored due to path filters (1)
site/static/notebooks/corpus.zipis excluded by!**/*.zip
📒 Files selected for processing (4)
site/content/contextual-data-platform/tutorials/_index.mdsite/content/contextual-data-platform/tutorials/evaluate-locally.mdsite/content/contextual-data-platform/tutorials/notebook-tutorial.mdsite/static/notebooks/Autograph_DEMO.ipynb
|
NOTE: Resources will be removed after review |
Co-authored-by: Paula Mihu <97217318+nerpaula@users.noreply.github.com>
Co-authored-by: Paula Mihu <97217318+nerpaula@users.noreply.github.com>
There was a problem hiding this comment.
I do have a general comment:
Should we merge most of this Tutorial actually into the notebook?
Realistically I would not switch between tutorial and notebook server back and forth.
The information is highly overlapping.
I do like the level the tutorial is written at way better then the Notebook, it hits more on the "why i am doing things" rather then just doing it.
So my suggestion:
Keep the "executable cells" in the notebook.
But replace most of the text with what has been written in the tutorial.
The Tutorial file should explain how to start a notebook server, click on it, upload everything you will need.
And then it should hand over to the Notebook alone driving the tutorial.
Also please change the focus form "Super generic" to "solve exactly the one corpus example".
Edit: This would also mean inside the Notebook, let us focus to have one run through, explaining the Variants we actually use there. Right now it explains in many places alternatives and all things one could do.
I would remove them for the "Tutorial" Notebook.
It should not be an API Reference description.
| weight: 18 | ||
| description: >- | ||
| A hands-on tutorial that takes you from a folder of documents to a queryable | ||
| Context Graph, driving the full AutoGraph pipeline from a Jupyter notebook |
There was a problem hiding this comment.
We should double check if "AutoGraph" is in-line with official naming still.
https://arangodb.atlassian.net/wiki/spaces/TUP/pages/3652780035/AutoGraph+Studio+Naming+2026-07-21
If I see this correctly this may be the "AutoGraph Studio" pipeline.
AFAIK AutoGraph stops at "you have a graph"
AutoRAG starts at "you want to ask the graph"
Both combined is AutoGraph Studio?
There was a problem hiding this comment.
This comment applies to more than one mention in this Document
| menuTitle: Notebook Tutorial | ||
| weight: 18 | ||
| description: >- | ||
| A hands-on tutorial that takes you from a folder of documents to a queryable |
There was a problem hiding this comment.
I would wish for a short paragraph purely on the problem space here:
Why should I do this tutorial?
What is the value I would get out of the Product?
Try to keep this low tech and non-arango specific.
I am not doing this tutorial because I want to run some commands and python code.
I am doing this because I have problem to solve, and I want to understand if this Product helps me solve the problem.
| and the retriever we deploy with AutoRAG and finally chat with. | ||
|
|
||
| This is the same workflow you would run in | ||
| [AutoGraph Studio](../../agentic-ai-suite/autograph/web-interface.md), the |
There was a problem hiding this comment.
Ah here we mention AutoGraph Studio.
Is Studio only the UI for the Flow?
How is the Flow called then, if AutoGraph is one step of it only?
|
|
||
| By the end, you will have: | ||
|
|
||
| - An AutoGraph service deployed on the platform, configured with your LLM. |
There was a problem hiding this comment.
| - An AutoGraph service deployed on the platform, configured with your LLM. | |
| - An AutoRAG service deployed on the platform, configured with your LLM. Allowing you to have an Agent reason about your documents. |
| By the end, you will have: | ||
|
|
||
| - An AutoGraph service deployed on the platform, configured with your LLM. | ||
| - Your documents uploaded through the File Manager and embedded into a Corpus Graph. |
There was a problem hiding this comment.
Do i as user really care about this?
my goal is to ask Questions.
IMO I would remove this entry
| - Your documents uploaded through the File Manager and embedded into a Corpus Graph. |
| PASSWORD = "<your-password>" | ||
| DB_NAME = "your-database" | ||
| PROJECT_NAME = "your-autograph-project" | ||
| LLM_API_KEY = "sk-..." |
There was a problem hiding this comment.
The needs to stay, please do not fill out one of our OpenAI keys ;)
| own values: | ||
|
|
||
| ```sh | ||
| SERVER_URL = "https://<EXTERNAL_ENDPOINT>:8529" |
There was a problem hiding this comment.
My suggestion:
Comment this out and say "only use this if you run on an external Notebook server"
Then we need to change the notebook to set SERVER_URL = os.environ["ARANGODB_ENDPOINT"] before loading the environment file
|
|
||
| | Variable | Purpose | | ||
| |---|---| | ||
| | `SERVER_URL` | Base URL of your platform gateway (port `8529`). | |
There was a problem hiding this comment.
Please only set this if running an external Jupiter server.
| |---|---| | ||
| | `SERVER_URL` | Base URL of your platform gateway (port `8529`). | | ||
| | `USERNAME` / `PASSWORD` | Platform credentials used to obtain the access token. | | ||
| | `DB_NAME` | The ArangoDB database that holds the project, documents, and Context Graph. | |
There was a problem hiding this comment.
| | `DB_NAME` | The ArangoDB database that holds the project, documents, and Context Graph. | | |
| | `DB_NAME` | The ArangoDB database where we want to store the Context Graph| |
| | `SERVER_URL` | Base URL of your platform gateway (port `8529`). | | ||
| | `USERNAME` / `PASSWORD` | Platform credentials used to obtain the access token. | | ||
| | `DB_NAME` | The ArangoDB database that holds the project, documents, and Context Graph. | | ||
| | `PROJECT_NAME` | The GenAI project name. It becomes the prefix for all collections AutoGraph creates (for example, `your-project_sources`, `your-project_domains`). | |
There was a problem hiding this comment.
Please use the above selected name for examples here.
|
Okay, commenting on Notebooks is hard. And GitHub kills the Python code, by rendering it as markdown. I will provide more feedback over slack. |
Co-authored-by: Michael Hackstein <michael@arangodb.com>
3 participants
Description
Upstream PRs
Summary by CodeRabbit