4itworks/Livvie Code Review
AI code review with multi-perspective reviewers, native suggestion blocks, and REQUEST_CHANGES support.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Activelast commit Jul 13, 2026
- License
- MIT
Pinned Snippet
uses: 4itworks/livvie_code_review@e1024d2a5623dda7b0f0934dfc35e2e270a8a147 # v2tags can be moved; commit SHAs can't. why a SHA?
Inputs
| name | description | required | default |
|---|---|---|---|
| github-token | GitHub token for posting PR reviews. | yes | ${{ github.token }} |
| llm-api-key | API key for your LLM provider. This is the only sensitive input — store it as a GitHub Secret. | yes | — |
| llm-base-url | Base URL for OpenAI-compatible API. Not sensitive — set as a plain string. | no | https://openrouter.ai/api/v1 |
| model | Model name (e.g. z-ai/glm-5.2, anthropic/claude-sonnet-4). Not sensitive — set as a plain string. | yes | — |
| review-instructions-file | Repository file with extra review instructions. | no | .github/code-reviewer.md |
| max-diff-size | Maximum diff size in characters per file. | no | 50000 |
| max-output-tokens | Maximum tokens for the LLM response (includes reasoning tokens if reasoning is enabled). | no | 16000 |
| reasoning-effort | Reasoning effort level for models that support it. Options: none, low, medium, high, max. Ignored by models that don't support reasoning. | no | none |
| fallback-model | Fallback model if the primary model fails after all retries. Set to empty to disable. | no | "" |
| request-changes-on-high | Post REQUEST_CHANGES when high-severity findings exist. | no | true |
| always-request-changes | (Deprecated) Use request-changes-on instead. | no | false |
| max-comments | Maximum inline comments to post. | no | 25 |
| include-severities | Comma-separated list of severities to include in the review. Options: low, medium, high. | no | low,medium,high |
| include-confidences | Comma-separated list of confidences to include in the review. Options: low, medium, high. | no | low,medium,high |
| request-changes-on | Comma-separated list of severities that trigger REQUEST_CHANGES. Use 'low,medium,high' to always request changes. | no | high |
| respect-author-replies | When true, the bot reads replies to its inline comments and dismisses findings whose author's justification is reasonable. | no | false |
| ignore-patterns | Comma-separated glob patterns for files to skip (e.g. *.g.dart,build/). Defaults to generated file patterns. | no | build/**,dist/**,node_modules/** |
| max-batches | Maximum number of file batches. Each batch is reviewed by every perspective, so total LLM calls = batches × perspectives. 0 = no limit. | no | 0 |
| context-window | Context window size in tokens for the model. Used for token budget calculation. Override if your model has a different context window. | no | 128000 |
| cross-file-budget-max | Maximum tokens to allocate for cross-file context in each batch. | no | 2000 |
| cross-file-budget-ratio | Percentage of context window (0-100) to allocate for cross-file context. | no | 5 |
| safety-margin | Additional tokens reserved for message wrappers and response overhead. | no | 500 |
| circuit-breaker-threshold | Consecutive failures before opening the circuit breaker for a model. | no | 3 |
| agents-dir | Directory containing agent .md files that define review perspectives. Each .md file = one reviewer. See README for file format and examples. | no | .github/livvie_code_review_agents |
| verbose | Log LLM reasoning traces and detailed debug info to the Actions log. | no | false |
Outputs
| name | description |
|---|---|
| review-id | The ID of the posted GitHub review. |
| finding-count | Total number of findings in the review. |