Skip to content

adapt to firecrest 2.5 and bump dependencies #26

adapt to firecrest 2.5 and bump dependencies

adapt to firecrest 2.5 and bump dependencies #26

Workflow file for this run

name: LLM PR Review
permissions:
contents: read
pull-requests: write
issues: write
on:
pull_request:
types: [opened, reopened, ready_for_review]
issue_comment:
env:
CRATE_NAME: coman
GITHUB_TOKEN: ${{ github.token }}
RUST_BACKTRACE: 1
RUSTFLAGS: "-Dwarnings"
jobs:
review:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
with:
fetch-depth: 0
- name: Get Access Token
env:
SERVICE_ACCOUNT_NAME: ${{ secrets.SERVICE_ACCOUNT_NAME}}
SERVICE_ACCOUNT_PASSWORD: ${{ secrets.SERVICE_ACCOUNT_PASSWORD}}
run: |
export SDSC_VLLM_TOKEN=$(curl -X POST https://authentik-server-runai-sharedllm-ralf.inference.compute.datascience.ch/application/o/token/ \
-H "Content-Type: application/x-www-form-urlencoded" \
-d "grant_type=client_credentials&client_id=P8dW2vrNPDa8d43qd4BK49eEDYJFtvYk&username=$SERVICE_ACCOUNT_NAME&password=$SERVICE_ACCOUNT_PASSWORD"\
|jq '.access_token' | tr -d '"')
echo "SDSC_VLLM_TOKEN=$SDSC_VLLM_TOKEN" >> "$GITHUB_ENV"
- name: PR Agent action step
id: pragent
uses: qodo-ai/pr-agent@main
env:
OPENAI_KEY: ${{ env.SDSC_VLLM_TOKEN }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
config__model: "openai/Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
config__fallback_models: '["openai/Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"]'
config__custom_model_max_tokens: "32000"
config__ai_timeout: "1800"
openai__extra_body: '{"chat_template_kwargs":{"enable_thinking":false}}'
OPENAI__API_BASE: "https://vllm-gateway-runai-sharedllm-ralf.inference.compute.datascience.ch/v1"
config__verbosity_level: "2"
github_action_config__auto_review: "true"
github_action_config__auto_describe: "true"
github_action_config__auto_improve: "true"
# note: this is a bit too chatty, with lots of comments like "X was changed, please make sure this is what you want"
# - name: AI-Review Step
# uses: Nikita-Filonov/ai-review@v0.51.0
# with:
# review-command: run-inline
# env:
# # --- LLM configuration ---
# LLM__PROVIDER: "OPENAI"
# LLM__META__MODEL: "Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
# LLM__META__MAX_TOKENS: "16000"
# LLM__HTTP_CLIENT__API_URL: "https://vllm-gateway-runai-sharedllm-ralf.inference.compute.datascience.ch/v1"
# LLM__HTTP_CLIENT__API_TOKEN: ${{ env.SDSC_VLLM_TOKEN }}
# # --- GitHub integration ---
# VCS__PROVIDER: "GITHUB"
# VCS__PIPELINE__OWNER: ${{ github.repository_owner }}
# VCS__PIPELINE__REPO: ${{ github.event.repository.name }}
# VCS__PIPELINE__PULL_NUMBER: ${{ github.event.number}}
# VCS__HTTP_CLIENT__API_URL: "https://api.github.com"
# VCS__HTTP_CLIENT__API_TOKEN: ${{ secrets.GITHUB_TOKEN }}
# REVIEW__IGNORE_CHANGES: '["Cargo.lock"]'