mingzaily/Issue AI Analyze

AI-assisted triage for GitHub issues with label mapping, duplicate hints, and structured comments.

Trust Signals

workflow.ymlSHA-pinned

uses: mingzaily/issue-ai-analyze@4819cc1bc46da45f73e85bba7821d793c65d8d58 # v1.0.0

tags can be moved; commit SHAs can't. why a SHA?

name	description	required	default
github-token	GitHub token with issue write permission. In most workflows this is just secrets.GITHUB_TOKEN.	yes	—
prompt-file	Optional custom prompt YAML file. Defaults to prompts/general.prompt.yml.	no	""
model	Optional model override written into the resolved prompt before inference.	no	""
language	Output language for bundled prompt instructions and built-in comments. Supported values: zh, en.	no	zh
label-map	Optional inline label mapping. Use one key=value entry per line, for example bug=type/bug. Use rerun=ai-rerun,ai-recheck for rerun labels.	no	""
label-map-file	Optional YAML file path for label management config. Supported keys are bug, question, enhancement, documentation, duplicate, needs-info, and rerun. Takes precedence over label-map.	no	""
openai-compatible-endpoint	Optional OpenAI-compatible endpoint. When set together with the token, GitHub Models is skipped.	no	""
openai-compatible-token	Optional token for the OpenAI-compatible endpoint.	no	""
openai-compatible-headers	Optional custom headers forwarded to actions/ai-inference.	no	""
comment-marker	Hidden marker used to find the latest AI analysis comment.	no	<!-- issue-ai-analyze -->
recent-comments-limit	Number of recent comments passed to the model.	no	10
open-issues-limit	Number of open issues included for duplicate detection.	no	50

name	description
should-run	Whether the current event triggered an analysis run.
ok	Whether AI normalization succeeded.
result-json	Normalized analysis result JSON.
labels	JSON array of labels applied to the issue after mapping.
category	Normalized category selected by the model.
disposition	Normalized disposition selected by the model.
needs-info	Whether the issue still needs more information after normalization.
comment-id	The AI comment ID created or updated during this run.
comment-strategy	Whether this run replaced the latest AI comment or created a new one.
transport	Which inference transport was selected.
resolved-model	Effective model resolved from the prompt file or action defaults.
resolved-response-format	Effective response format resolved from the prompt file.
resolved-model-parameters	Effective modelParameters object resolved from the prompt file, serialized as JSON.