Chore: Use auto-triager github action directly and pass labels and prompts files (#98602)

* Add auto triager files and use them in the auto-triager github action

* add auto-triager file to codeowners
This commit is contained in:
Esteban Beltran 2025-01-07 15:04:20 +01:00 committed by GitHub
parent 976cfcbf4f
commit 92fb6e765f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
5 changed files with 186 additions and 8 deletions

1
.github/CODEOWNERS vendored
View File

@ -746,6 +746,7 @@ embed.go @grafana/grafana-as-code
/.github/pr-checks.json @tolzhabayev
/.github/pr-commands.json @tolzhabayev
/.github/renovate.json5 @grafana/frontend-ops
/.github/workflows/auto-triager/ @grafana/plugins-platform-frontend
/.github/workflows/alerting-swagger-gen.yml @grafana/alerting-backend
/.github/workflows/auto-milestone.yml @grafana/grafana-developer-enablement-squad
/.github/workflows/backport.yml @grafana/grafana-developer-enablement-squad

View File

@ -0,0 +1,125 @@
area/admin/user
area/alerting
area/annotations
area/auth
area/auth/ldap
area/auth/oauth
area/auth/rbac
area/auth/serviceaccount
area/backend
area/backend/api
area/backend/db
area/backend/db/migration
area/backend/db/mysql
area/backend/db/postgres
area/backend/db/sql
area/backend/db/sqlite
area/configuration
area/dashboard/annotations
area/dashboard/data-links
area/dashboard/edit
area/dashboard/folders
area/dashboard/import
area/dashboard/kiosk
area/dashboard/links
area/dashboard/rows
area/dashboard/scenes
area/dashboard/settings
area/dashboard/snapshot
area/dashboard/templating
area/dashboard/timerange
area/dashboard/tv
area/dashboard/variable
area/dashboards/panel
area/data/export
area/editor
area/explore
area/exploremetrics
area/expressions
area/field/overrides
area/frontend/library-panels
area/frontend/login
area/image-rendering
area/internationalization
area/legend
area/library-panel
area/navigation
area/panel/annotation-list
area/panel/barchart
area/panel/bargauge
area/panel/candlestick
area/panel/canvas
area/panel/dashboard-list
area/panel/edit
area/panel/edit
area/panel/field-override
area/panel/flame-graph
area/panel/gauge
area/panel/geomap
area/panel/heatmap
area/panel/histogram
area/panel/logs
area/panel/node-graph
area/panel/node-graph
area/panel/piechart
area/panel/repeat
area/panel/singlestat
area/panel/stat
area/panel/state-timeline
area/panel/status-history
area/panel/table
area/panel/timeseries
area/panel/traceview
area/panel/trend
area/panel/xychart
area/permissions
area/playlist
area/plugins
area/plugins-catalog
area/provisioning
area/provisioning/datasources
area/public-dashboards
area/query-library
area/recorded-queries
area/scenes
area/search
area/security
area/streaming
area/templating/repeating
area/tooltip
area/transformations
datagrid
datasource/Alertmanager
datasource/Azure
datasource/azure-cosmosdb
datasource/BigQuery
datasource/CloudWatch
datasource/CloudWatch Logs
datasource/CSV
datasource/Elasticsearch
datasource/GitHub
datasource/GoogleCloudMonitoring
datasource/GoogleSheets
datasource/grafana-pyroscope
datasource/Graphite
datasource/InfluxDB
datasource/Jaeger
datasource/JSON
datasource/Loki
datasource/MSSQL
datasource/MySQL
datasource/OpenSearch
datasource/OpenTSDB
datasource/Parca
datasource/Phlare
datasource/Postgres
datasource/Prometheus
datasource/SiteWIse
datasource/Splunk
datasource/Tempo
datasource/TestDataDB
datasource/Timestream
datasource/X-Ray
datasource/Zabbix
datasource/Zipkin
team/grafana-aws-datasources

View File

@ -0,0 +1,25 @@
You are an expert Grafana issues categorizer.
You are provided with a Grafana issue. Your task is to categorize the issue by analyzing the issue title and description to determine the most relevant category and type from the provided lists. Focus on precision and clarity, selecting only the most pertinent labels based on the issue details. Ensure that your selections reflect the core problem or functionality affected.
The output should be a valid JSON object with the following fields:
* id (string): The ID of the current issue.
* categoryLabel (array of strings): The category labels for the current issue, emphasizing key terms and context.
* typeLabel (array of strings): The type of the current issue, emphasizing clarity and relevance.
**Instructions**:
1. **Contextual Analysis**: Understand the context and intent behind the issue description. Analyze the overall narrative and relationships between different components within Grafana. Consider dependencies and related components to inform your decision.
2. **Category and Type Differentiation**: Use language cues and patterns to differentiate between similar categories and types. Provide examples and counterexamples to clarify distinctions. Prioritize primary components over secondary ones unless they are critical to the issue.
3. **Historical Data Utilization**: Compare current issues with past resolved issues by analyzing similarities in problem descriptions, leveraging patterns to inform categorization. Use historical data to recognize patterns and inform your decision-making.
4. **Confidence Scoring**: Implement a confidence scoring mechanism to flag issues for review if the confidence is below a predefined threshold. Clearly indicate thresholds for high and low confidence predictions. Provide clarifying questions if data is ambiguous.
5. **Feedback Loop Integration**: Integrate feedback from incorrect predictions to refine understanding and improve future predictions. Conduct error analysis to identify patterns in misclassifications and adapt your approach accordingly.
6. **Semantic Analysis**: Evaluate the underlying intent of the issue using semantic analysis, considering broader implications and context. Leverage metadata or historical patterns to improve accuracy.
7. **Avoid Over-Specification**: Maintain precision and conciseness, avoiding unnecessary details. Prioritize clarity and flag for further review if uncertain.
8. **Consistent JSON Formatting**: Ensure the output maintains a consistent JSON structure with uniform formatting for readability and scalability.
**Next Steps and Insights**:
- Suggest potential next steps or resources that could help address the issue, providing actionable insights to enhance user engagement.
- Regularly test responses against edge cases to ensure robustness and adaptability.
- Stay updated with changes in category and type lists to remain current.
Provide a brief explanation of the categorization decision, highlighting key terms or context that influenced the choice. Use user-centric language and technical details to ensure the explanation is comprehensive and insightful.

View File

@ -0,0 +1,30 @@
type/accessibility
type/angular-2-react
type/browser-compatibility
type/bug
type/build-packaging
type/chore
type/ci
type/cleanup
type/codegen
type/community
type/debt
type/design
type/discussion
type/docs
type/duplicate
type/e2e
type/epic
type/feature-request
type/feature-toggle-enable
type/feature-toggle-removal
type/performance
type/poc
type/project
type/proposal
type/question
type/refactor
type/regression
type/roadmap
type/tech
type/ux

View File

@ -82,23 +82,20 @@ jobs:
app_id: ${{ env.GH_APP_ID }}
private_key: ${{ env.GH_APP_PEM }}
- name: Checkout auto-triager repository
- name: Checkout
uses: actions/checkout@v4
with:
repository: grafana/auto-triager
path: auto-triager
token: ${{ steps.generate_token.outputs.token }}
- name: Send issue to the auto triager action
id: auto_triage
# https://github.com/grafana/auto-triager/blob/main/action.yml
#uses: grafana/auto-triager@main
uses: ./auto-triager
uses: grafana/auto-triager@main
with:
token: ${{ steps.generate_token.outputs.token }}
issue_number: ${{ github.event.issue.number }}
openai_api_key: ${{ env.AUTOTRIAGER_OPENAI_API_KEY }}
add_labels: true
labels_file: ${{ github.workspace }}/.github/auto-triager/labels.txt
types_file: ${{ github.workspace }}/.github/auto-triager/types.txt
prompt_file: ${{ github.workspace }}/.github/auto-triager/prompt.txt
- name: Labels from auto triage
run: |