Files

T

werkator build succeeded after 01:57

Details

werkrevisor 0.8.0: repository config layer, clear context reasons, docs only as diff (#5 )

werkrevisor 0.8.0: repository config layer, clear context reasons, docs only as diff.

- Config layers: built-in defaults < host < `.werkrevisor.toml` read from the PR base; the repository may only tighten cost and model limits, and `.werkrevisor.toml` is always a risky path.
- One clear reason per cause instead of one line per file.
- Documentation files go to the model only as diff and are judged against the writing style in AGENTS.md.
- AGENTS.md with the shared writing style.

Details: the PR document under `docs/prs/`.

🤖 Generated with [Claude Code](https://claude.com/claude-code)
Reviewed-on: #5
Co-authored-by: claude-code <claude-code@javagil.de>

2026-10-05 16:00:59 +02:00

1.9 KiB

Raw Permalink Blame History

werkrevisor Agent Instructions

Project

werkrevisor is an AI review gate for Gitea pull requests, written in Python. It applies deterministic rules, asks a model through PR-Agent 0.47.0 for a review, and can ask the classifier Jev about the PR documents. Werkator runs it as a pull_request_target-style build; its exit code is the merge gate. See README.md for the verdicts, the configuration and the operation.

Build and Test

werkrevisor uses only the Python standard library and the packages in the pinned PR-Agent image. Run the tests inside that image, from the repository root:

docker run --rm --network none -v "$PWD":/werkrevisor -w /werkrevisor --entrypoint python \
    pragent/pr-agent:0.47.0@sha256:7d98954c29289846a08faf00aaea0f1f1375b1676cd8a2ed303e0af07010af7a -m unittest -v

The tests make no network calls: no LLM, no Jev, no Gitea.

Conventions

Comments explain why, never what or how; default to no comment.
Markdown: one sentence per line.
German docs (docs/*.md) avoid anglicisms in the prose; code identifiers and technical terms stay as they are.
Every PR has a PR document under docs/prs/; follow the workflow in docs/prs/README.md.
.werkrevisor.toml holds the repository part of the review config; see the README.

Writing Style

Be terse; omit obvious details, implementation narration, and summaries. Prefer one short sentence per change. Remove all mannered prose; use literal language. Use ASD-STE100 as guidance: simple words, active voice, consistent terminology. Preserve code identifiers and established technical terms.

Write at most one sentence per line. Wrap sentences longer than 128 characters at sensible points, preferably after commas, colons, semicolons or before 'and' or 'or'. Paragraphs should comprise at most 5 sentences; add blank lines between paragraphs. Format enumerations as such instead of clunky lists.