feat(S01): add ci-pr-review workflow and RULESET.md org setup guide#1
Open
feat(S01): add ci-pr-review workflow and RULESET.md org setup guide#1
Conversation
- ci-pr-review.yml: two-job workflow with Claude Opus 4.6 + extended thinking, auto-review/describe/improve, slash commands, draft/bot exclusion (R001-R007) - RULESET.md: org admin guide for secret and ruleset setup (R008)
qodo-ai/pr-agent@main is a Docker action — it doesn't accept 'with:' inputs other than 'entryPoint'/'args'. GITHUB_TOKEN must be passed as an env var.
|
Failed to generate code suggestions for PR |
… 4.6
claude-opus-4-6 rejects thinking:{type:enabled,budget_tokens:N} — this API
is deprecated on 4.6 models. Opus 4.6 uses adaptive thinking by default at
high effort, which means it almost always thinks deeply without explicit
budget configuration. Removing config.enable_claude_extended_thinking and
config.extended_thinking_budget_tokens.
|
Failed to generate code suggestions for PR |
Anthropic requires temperature=1 when reasoning/thinking is active. pr-agent defaults to temperature=0.2 + reasoning_effort=medium (adaptive thinking on 4.6 models) which triggers a 400 from the API.
|
Failed to generate code suggestions for PR |
…g_effort injection pr-agent auto-injects reasoning_effort=medium for Anthropic models, which LiteLLM translates to thinking params — causing a malformed request that Anthropic rejects with a vague 400. custom_reasoning_model=true tells pr-agent to treat the model as a custom/non-reasoning model and skip the reasoning_effort injection, letting the API call go through cleanly.
|
Failed to generate code suggestions for PR |
…ing injection
pr-agent auto-sets reasoning_effort=medium for Anthropic models even with
custom_reasoning_model=true in some versions. LiteLLM translates this to
thinking:{type:adaptive} which Opus 4.6 requires temp=1 for — but the
combination of params it sends causes a malformed 400. Setting
reasoning_effort='' disables the injection entirely; Opus 4.6 still uses
adaptive thinking by default without explicit params.
|
Failed to generate code suggestions for PR |
Replace qodo-ai/pr-agent (requires Anthropic API key) with Anthropic's official claude-code-action which supports Claude Code OAuth tokens from Max subscriptions. No separate API billing needed. - Auto-review on PR open/reopen/sync/ready-for-review with structured output - @claude slash commands in PR comments - Claude Opus 4.6 model, draft/bot exclusion, concurrency control - Updated RULESET.md for new secret setup (CLAUDE_CODE_OAUTH_TOKEN)
claude-code-action@v1 doesn't have a 'model' input — pass via claude_args.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds org-wide PR review automation via qodo-ai/pr-agent with Claude Opus 4.6.
Changes
.github/workflows/ci-pr-review.ymlTwo-job workflow:
pr_agent_review— fires on PR open/reopen/sync/ready-for-review, skips drafts and bots, posts review + description + improvement comments automaticallypr_agent_commands— fires on issue comments in PRs for slash commands (e.g./review,/describe)Model config:
anthropic/claude-opus-4-6with extended thinking (10240 token budget)anthropic/claude-sonnet-4-6ANTHROPIC_KEYorg secret (ANTHROPIC.KEYenv var — dot notation required by pr-agent)RULESET.mdStep-by-step guide for org admins to:
ANTHROPIC_KEYorg secretRequirements addressed