wordmill
A work-in-progress.
wordmill is a document summary API service. It provides a simple REST API which accepts requests to summarize a document. Under the hood, it reaches out to a LLM hosted on any OpenAI-compatible API service.
It is designed to abstract away the "AI" details from your user base. Many people in your organization want to summarize documents. Not as many of them care to know all the details related to LLMs, prompts, document prep, model selection, etc. The service will allow administrators to configure and customize these aspects based on the incoming document type. All the end users need to do is request a summary.
Setup
-
It is recommended to install pyenv.
NOTE: Make sure to follow all steps! (A, B, C, D, and so on)
-
Set up the virtual environment:
pipenv install --dev
-
Add your LLM access info into
.env, example:LLM_API_KEY=<your key> LLM_BASE_URL="https://my-llm-server:443/v1" LLM_MODEL_NAME="mistral-7b-instruct"
Running API server
To run the server:
pipenv shell
flask run
Example Usage of the API server
The service accepts a request to summarize the document and returns a URL that you should visit to check the status of your summary.
A background task reaches out to the LLM and awaits the response. Eventually, the status of your summarize task will shift to 'done' and you can view the LLM-generated content. Your task may also shift to 'error' if something went wrong.
Since these summaries do not need to be long-lived, currently we are using flask-caching's "SimpleCache" to store the data. For production purposes, the cache service used by flask-caching will need to be changed to redis or memcached
import json
import requests
import time
# load the document you wish to summarize
with open("incident.json") as fp:
data = json.load(fp)
# customize the prompt passed to the LLM (optional)
requests.post(
"http://127.0.0.1:8000/prompt",
json={"prompt": "Please summarize this document:\n\n{document}"}
)
# submit request to summarize and get background task id
id = requests.post("http://127.0.0.1:8000/summarize", json={"document": data}).json()["id"]
# repeatedly check on the 'summarize' task and wait for summary to be generated...
while True:
time.sleep(5)
summary = requests.get(f"http://127.0.0.1:8000/summary/{id}").json()
if summary["status"] == "done":
print(summary["content"])
break
Metadata
Release files for wordmill 0.0.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| wordmill-0.0.2.tar.gz | 41.5 kB | With provenance | Details for wordmill-0.0.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| wordmill-0.0.2-py3-none-any.whl | Python 3 | none | any | Details for wordmill-0.0.2-py3-none-any.whl |
Total release size: 51.4 kB
Release files / wordmill-0.0.2.tar.gz
| Download URL | wordmill-0.0.2.tar.gz |
|---|---|
| Size | 41.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d1c5c7b846da5ab908f456216f88a6a7e3ce67ed2d7f7b91d5ad64ff7c09c416
|
|
BLAKE2b-256 checksum How to use checksums |
9e635ffa10266a87622205bf40ad06068138e6d52cc740c46c09e14fce328522
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 6, 2025.
Transparency logRelease files / wordmill-0.0.2-py3-none-any.whl
| Download URL | wordmill-0.0.2-py3-none-any.whl |
|---|---|
| Size | 9.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
fa20cbb5bc1417dcd4d02bfc203fe2b02702db3d424224ce3adc362c13d6a579
|
|
BLAKE2b-256 checksum How to use checksums |
b508acf700b42b72cf3f4e397a72a56a5770e1226cd3877bca3a91cf7c58250e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 6, 2025.
Transparency logSecurity for wordmill 0.0.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for wordmill-0.0.2.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 6, 2025.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.