This release is a pre-release and may not be stable for production use.
ratel-ai-rankers-plugin
Decision-model rankers as retrieve/rerank plugins for ratel-ai, the Ratel Python SDK.
The SDK ranks through functions you give it (retrieve_fn, reranker_fn; ADR 0027) and knows no vendor's API. This package holds the vendor clients, so the SDK stays vendor-free and a new vendor ships without an SDK release (ADR 0028).
pip install ratel-ai ratel-ai-rankers-plugin
Plugins
Each returns a RetrieverPlugin with .retrieve (for retrieve_fn, method="custom") and .rerank (for reranker_fn). Every failure raises the SDK's RetrieverError. When a reranker fails with transient=True, the search keeps the first stage's order.
from ratel_ai import ToolCatalog
from ratel_ai_rankers_plugin import ratel_jev_plugin, ratel_openai_decision_plugin
jev = ratel_jev_plugin() # key in TYPESAFE_API_KEY
ToolCatalog(method="custom", retrieve_fn=jev.retrieve) # Jev ranks every tool
ToolCatalog(method="bm25", reranker_fn=jev.rerank) # Jev reranks BM25's top 50
decision = ratel_openai_decision_plugin() # key in OPENAI_API_KEY; public beta
ToolCatalog(method="bm25", reranker_fn=decision.rerank, reranker_depth=20)
ratel_jev_plugin(url=None, api_key_env=None, model=None): Jev (TypeSafe AI). Defaults:https://api.typesafe.ai,TYPESAFE_API_KEY,jev-latest.ratel_openai_decision_plugin(url=None, api_key_env=None, model=None): OpenAI's Decisions API (public beta). Defaults:https://api.openai.com,OPENAI_API_KEY,gpt-6-luna. The first plugin made emits a one-timeExperimentalWarning; setRATEL_EXPERIMENTAL_SILENCE=1to silence it. A model refusal iscode="Refused", which is transient.
Both plugins send the query and each candidate's searchable text to the vendor. Above 150 candidates they rank groups in parallel and put the group winners into one final question. Picks below probability 0.01 are dropped, except the best one.
Writing your own plugin
The plugin contract belongs to the SDK. This package re-exports it so a plugin can import everything from one place:
from ratel_ai_rankers_plugin import RankCandidate, RankedId, define_plugin
async def rank(query: str, candidates: list[RankCandidate], top_k: int) -> list[RankedId]:
...
my_plugin = define_plugin(retrieve=rank, rerank=rank)
choice_ranker_plugin(ranker) turns any object with a rank(query, [(id, text)], top_k, kind) method into a plugin.
Layout
native/ pyo3 binding to the ratel-ai-rankers crate (../core)
ratel_ai_rankers_plugin/ the plugins, the shared adapter, type stubs
tests/ plugin tests against local vendor stand-ins
Build & test
From this directory, with uv:
uv venv --python 3.11 .venv
uv pip install --python .venv maturin pytest pytest-asyncio ruff mypy -e ../../telemetry/python
unset CONDA_PREFIX
# The in-repo SDK, then this package; --no-deps because the SDK's version is only
# bumped at release, so the pin on the next SDK may not be met yet.
uv pip install --python .venv -e ../../sdk/python
uv pip install --python .venv --no-deps -e .
.venv/bin/ruff check . && .venv/bin/mypy ratel_ai_rankers_plugin && .venv/bin/pytest
Released as the rankers-py-v* unit (ADR-0008).
Metadata
Release files for ratel-ai-rankers-plugin 0.1.0rc1
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 | |
|---|---|---|---|---|
| ratel_ai_rankers_plugin-0.1.0rc1.tar.gz | 48.8 kB | With provenance | Details |
Built distributions (wheels)
| File | Reset | |||
|---|---|---|---|---|
| ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-win_amd64.whl | CPython 3.9 | abi3 | Windows x86-64 | Details |
| ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl | CPython 3.9 | abi3 | Linux glibc 2.17+ x86-64 | Details |
| ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl | CPython 3.9 | abi3 | Linux glibc 2.17+ ARM64 | Details |
| ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_11_0_arm64.whl | CPython 3.9 | abi3 | macOS 11.0+ ARM64 | Details |
| ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_10_12_x86_64.whl | CPython 3.9 | abi3 | macOS 10.12+ x86-64 | Details |
Total release size: 6.2 MB
Release files / ratel_ai_rankers_plugin-0.1.0rc1.tar.gz
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1.tar.gz |
|---|---|
| Size | 48.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
a22f55b74c73cb98b3ff9f4e324f7c987697adfed08bc30fa9f0111a37c135ad
|
|
BLAKE2b-256 checksum How to use checksums |
098407eb7626392ac03117707be60aef01b6cc32061edbc46d86453e7e7e134b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logRelease files / ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-win_amd64.whl
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-win_amd64.whl |
|---|---|
| Size | 1.2 MB |
| Tags | CPython 3.9 Windows x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
796f339cdd94929bb9d4d5ee640133aa9b45606f507caac1be992a9b16003c44
|
|
BLAKE2b-256 checksum How to use checksums |
e7c42aaa00860fc5403f348b4ff4376e86e8d4e2afa10ea43d84846c3ac7568a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logRelease files / ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 1.3 MB |
| Tags | CPython 3.9 Linux glibc 2.17+ x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
f43e5f182772adc8c86adc4150ebbfc797fd09965ec983cb506cad0b82268be3
|
|
BLAKE2b-256 checksum How to use checksums |
07df4b1f71da36c64605dad3c07113fdcdf886c9714c9b44467d66f43e934529
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logRelease files / ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.3 MB |
| Tags | CPython 3.9 Linux glibc 2.17+ ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
60ddfa3d2c4128308b27cdf0876d04b5731e1ffad6a8ca3f40bb58ed33236dbc
|
|
BLAKE2b-256 checksum How to use checksums |
7ae3553af76092879cb312e4810d28962c777173e04d06dd284c37dc31e41870
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logRelease files / ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_11_0_arm64.whl
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.2 MB |
| Tags | CPython 3.9 abi3 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
e3b10cf0920d4a11f78f1a485bdd1076114456136fea8fde7c40ac40953936a0
|
|
BLAKE2b-256 checksum How to use checksums |
1edec6eb17203a8ab65ae5a7f74bdfdc4843a94f6a978c0a63fd2abcf01f33fc
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logRelease files / ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_10_12_x86_64.whl
| Download URL | ratel_ai_rankers_plugin-0.1.0rc1-cp39-abi3-macosx_10_12_x86_64.whl |
|---|---|
| Size | 1.2 MB |
| Tags | CPython 3.9 abi3 macOS 10.12+ x86-64 |
|
SHA-256 checksum How to use checksums |
aea632c63deb128c40d630c30cddc55792b8b6574f9b7e5a21b52d9cd3918484
|
|
BLAKE2b-256 checksum How to use checksums |
23981fe9e27d279becc45735237256ee7c23f7eb32b13c6010e3989f9fc780b0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
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 Oct 22, 2026.
Transparency logSecurity for ratel-ai-rankers-plugin 0.1.0rc1
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
All 6 files in this release have been uploaded with provenance metadata.
Attestation for ratel_ai_rankers_plugin-0.1.0rc1.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 Oct 22, 2026.
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.