Skip to main content

Deterministic prompt and rubric drift detection for PR review.

Project description

prompt-rubric-drift

prompt-rubric-drift compares prompt and rubric files across two directories and generates deterministic PR-review notes. It flags changed scoring boundaries, criteria additions/deletions, weight changes, stricter or looser wording, missing examples, prompt-injection exposure, and judge contract drift.

No model call is required.

Scope

This is not a model judge, policy approval, compliance review, or guarantee that a prompt or rubric change is safe. It is a deterministic PR-review aid for surfacing likely drift that still needs human review.

Quick start

python -m pip install -e .
prompt-rubric-drift compare examples/before examples/after --out report.md

The report is designed to be pasted as a PR comment or published by the bundled GitHub Action.

GitHub Action

steps:
  - uses: actions/checkout@v4
  - uses: auraoneai/prompt-rubric-drift@v0.1.0
    with:
      before: prompts/base
      after: prompts/head
      output: prompt-rubric-drift.md
      comment: "true"

When comment is enabled on a pull request event, the action writes the report to the job summary and creates or updates one deterministic PR comment. Grant issues: write or pull-requests: write permission in the workflow.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

prompt_rubric_drift-0.1.0.tar.gz (5.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

prompt_rubric_drift-0.1.0-py3-none-any.whl (6.9 kB view details)

Uploaded Python 3

File details

Details for the file prompt_rubric_drift-0.1.0.tar.gz.

File metadata

  • Download URL: prompt_rubric_drift-0.1.0.tar.gz
  • Upload date:
  • Size: 5.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for prompt_rubric_drift-0.1.0.tar.gz
Algorithm Hash digest
SHA256 208c2094fe90dbbd04665d20b7773ba11cdf884688fb4db8a0836f7801672a44
MD5 c2eca15cb2e656b3d262957267564f53
BLAKE2b-256 2e81b47a83442939e9c3c121a64f3271231408d104fc18ae1d6f1fcdc3f1982e

See more details on using hashes here.

Provenance

The following attestation bundles were made for prompt_rubric_drift-0.1.0.tar.gz:

Publisher: release-python.yml on auraoneai/prompt-rubric-drift

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file prompt_rubric_drift-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for prompt_rubric_drift-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 70123e920ab5b3abd4cccc26dec10b5e00bf83265332e0b670828273f6559d1a
MD5 41e23945129fdfca3c97c9d4f8fe122c
BLAKE2b-256 616922b55ca2b46f2cd2ed4db1cac27982658b0bbdc1aba56d1e3c6ecee21cd6

See more details on using hashes here.

Provenance

The following attestation bundles were made for prompt_rubric_drift-0.1.0-py3-none-any.whl:

Publisher: release-python.yml on auraoneai/prompt-rubric-drift

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page