User repository on GitHub
Google ADK vs Microsoft Agent Framework for structured-output agents, with TypeSafe Jev as a vendor-neutral QC gate
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Practices & Patterns
Jev Storyboard Lab
Companion code comparing Google ADK and Microsoft Agent Framework on one Pydantic VideoTimeline schema, with the same Jev client checking every generated segment for sane captions and durations.
- Stars
- 3
- Language
- Python
- License
- MIT
- Published by
- Community
- Author
- jimmyliao
- Last verified
- Oct 2026
About this project
Same Pydantic schema, two agent frameworks, one vendor-neutral QC gate. This repo is the companion code for a short article series comparing Google ADK and Microsoft Agent Framework for building a structured-output agent — and showing that TypeSafe AI's Jev (a calibrated decision API, not an LLM) can sit underneath either one unchanged as a scene-level QC gate.
Quoted from the project's README. Read the full README.
- Languages
- Python
Highlights
- One canonical Pydantic VideoTimeline schema across both demos, plus a discriminator-free Azure variant.
- Same Jev client call scores every segment whether the timeline came from Gemini or Azure OpenAI.
- Thin Jev client with live tests that skip cleanly when no TypeSafe key is set.
- Documents the ADK output_schema field name and the Azure oneOf rejection from live runs.
Watch out
The paid Gemini and Azure generation calls need your own keys and are not run in CI. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
Schema validation only checks shape. This repo adds a content check: Jev scores whether each segment caption plausibly fits inside its duration, catching nonsense that still validates.
Yes. The shared client and live tests under common/ only need a TypeSafe key. Each demo additionally needs its own provider key, so run whichever one you hold credentials for.
Azure structured outputs reject discriminated unions with a 400 oneOf error. The repo ships a discriminator-free VideoTimelineAzure for the wire format, then converts back to the canonical schema before QC.
The demo imports were checked against the installed SDKs at write time, including the ADK output_schema field name. Provider calls still need paid access, so pin versions and open an issue on SDK surface changes.
common/ only needs TYPESAFE_API_KEY, and its live tests skip cleanly without one; uv sync then uv run pytest tests/ -v exercises the same Jev client both demos use, so you can check the QC gate before wiring Gemini or Azure.
The ADK demo runs Google's gemini-3.1-pro-preview with output_schema, and the Microsoft Agent Framework demo runs Azure OpenAI with response_format; both feed a timeline into the same Jev segment check.
The Day 2 live end-to-end run produced 7 segments with 1 flagged, and because LLM generation is non-deterministic a rerun will not reproduce that same count.
More Jev resources like this
9GitHub stars
Research-memory skill that keeps experiment claims traceable with ten operations, provenance gates, and optional batched Jev audits over memory files.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
poiuyjie/jev_project_contextPython
poiuyjie/jev_project_contextPython11GitHub stars
Local memory layer that stores verbatim agent quotes in per-project SQLite and grades them with Jev into stale, wrong, unsupported, or keep verdicts.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
romiluz13/jevmoryPython
romiluz13/jevmoryPython20GitHub stars
Local harness that runs the open Laya decision model on your own machine and wires it into browser agents through a TypeSafe-compatible /v1/systemone dialect.
Jev Practices & Patterns#community#awesomejev#catalog-import
7GitHub stars
Python client and decision toolkit for the Jev System One endpoint with noul, choice, and score primitives, batched asks, composition helpers, concurrent evaluation, and an MCP server.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
docxology/daf-jevPython
docxology/daf-jevPython7GitHub stars
Independent open-source agent runtime where Jev makes typed tool decisions, an LLM authors text only when needed and reviews low-confidence calls, and code owns execution.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
parkavenue9639/jevloopPython
parkavenue9639/jevloopPython4GitHub stars
Coding agent that pairs explicit SQLite-typed context chunks with a small TypeSafe judgment model, so the large generation model only sees selected context.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
arczhi/jetPython
arczhi/jetPythonFrom the Jev guides
Original write-ups on what this entry is for.
Most Jev accuracy problems are state problems. What belongs in a state, what is a pre-made judgment to leave out, and why naming the slice you ask about matters more than the wording.
Jev Practices & Patterns#state#design#primitives
Read article
A Noul returns one number and no separate confidence. How to read the value, pin its boundary with criteria, split compound questions, and set a per-action threshold that matches the cost of being wrong.
Jev Practices & Patterns#noul#primitives#confidence
Read article
A Score is a weighted position, not a ruler. Why the number can land between levels, why levels are judged separately, and how to combine several atomic scores in code instead of one crowded rubric.
Jev Practices & Patterns#score#primitives#scoring
Read article
From the Jev community
Posts from builders shipping with Jev right now.