fix: use structured output for classify-issue-severity workflow #21240
Conversation
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
The claude-code-action's execution file format doesn't include the assistant's text response in an easily accessible field when not using --json-schema. The previous fix attempted to parse the execution file as Anthropic Messages API format, but claude-code uses its own format. This fix: 1. Adds --json-schema parameter to enable structured output 2. Uses the structured_output output directly from claude-code-action 3. Removes the complex jq parsing of the execution file The JSON schema enforces: - Required fields: status, reasoning - Conditional requirements based on status value - Valid enum values for status and severity Supersedes PR #21238 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
DevelopmentCats
approved these changes
Dec 11, 2025
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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
Fixes the extraction error in the
classify-issue-severity.ymlworkflow by using structured output instead of parsing the execution file.Root Cause
The claude-code-action uses its own stream-json format (not the Anthropic Messages API format). Without
--json-schema, the assistant's text response isn't available in a structured field in the execution file. The previous fix (#21238) tried to extract text fromcontentarrays, but that doesn't match the actual format.Solution
Use the claude-code-action's structured output feature:
Add
--json-schemaparameter with a JSON schema that matches the expected output formatstatus,reasoningstatusvaluestatusandseverityUse
structured_outputoutput directly from the claude-code-action--json-schemais providedSimplify extraction step
Changes
.github/workflows/classify-issue-severity.yml:claude_argswith--json-schemaparameterexecution_fileto usingstructured_outputVerification
The structured output approach is the recommended way to extract data from claude-code-action, as shown in their test workflows:
https://github.com/anthropics/claude-code-action/blob/main/.github/workflows/test-structured-output.yml
Supersedes
🤖 Generated with Claude Code