update: issue triage workflows tags duplicate issues (#5868)

This commit is contained in:
Gaurav 2025-08-11 11:48:57 -07:00 committed by GitHub
parent 239ba63d28
commit 6390b81646
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 250 additions and 0 deletions

View File

@ -28,6 +28,7 @@ permissions:
id-token: 'write'
issues: 'write'
statuses: 'write'
packages: 'read'
jobs:
triage-issue:
@ -201,3 +202,149 @@ jobs:
issue_number: '${{ github.event.issue.number }}',
body: 'There is a problem with the Gemini CLI issue triaging. Please check the [action logs](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}) for details.'
})
deduplicate-issues:
if: >
github.repository == 'google-gemini/gemini-cli' &&
vars.TRIAGE_DEDUPLICATE_ISSUES != '' &&
(github.event_name == 'issues' ||
github.event_name == 'workflow_dispatch' ||
(github.event_name == 'issue_comment' &&
contains(github.event.comment.body, '@gemini-cli /deduplicate') &&
(github.event.comment.author_association == 'OWNER' ||
github.event.comment.author_association == 'MEMBER' ||
github.event.comment.author_association == 'COLLABORATOR')))
timeout-minutes: 20
runs-on: 'ubuntu-latest'
steps:
- name: 'Checkout repository'
uses: 'actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683'
- name: 'Generate GitHub App Token'
id: 'generate_token'
uses: 'actions/create-github-app-token@df432ceedc7162793a195dd1713ff69aefc7379e'
with:
app-id: '${{ secrets.APP_ID }}'
private-key: '${{ secrets.PRIVATE_KEY }}'
- name: Log in to GitHub Container Registry
uses: docker/login-action@v3
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: 'Run Gemini Issue Deduplication'
uses: 'google-github-actions/run-gemini-cli@20351b5ea2b4179431f1ae8918a246a0808f8747'
id: 'gemini_issue_deduplication'
env:
GITHUB_TOKEN: '${{ steps.generate_token.outputs.token }}'
ISSUE_TITLE: '${{ github.event.issue.title }}'
ISSUE_BODY: '${{ github.event.issue.body }}'
ISSUE_NUMBER: '${{ github.event.issue.number }}'
REPOSITORY: '${{ github.repository }}'
FIRESTORE_PROJECT: '${{ vars.FIRESTORE_PROJECT }}'
with:
gcp_workload_identity_provider: '${{ vars.GCP_WIF_PROVIDER }}'
gcp_project_id: '${{ vars.GOOGLE_CLOUD_PROJECT }}'
gcp_location: '${{ vars.GOOGLE_CLOUD_LOCATION }}'
gcp_service_account: '${{ vars.SERVICE_ACCOUNT_EMAIL }}'
gemini_api_key: '${{ secrets.GEMINI_API_KEY }}'
use_vertex_ai: '${{ vars.GOOGLE_GENAI_USE_VERTEXAI }}'
use_gemini_code_assist: '${{ vars.GOOGLE_GENAI_USE_GCA }}'
settings: |-
{
"mcpServers": {
"issue_deduplication": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"--network", "host",
"-e", "GITHUB_TOKEN",
"-e", "GEMINI_API_KEY",
"-e", "DATABASE_TYPE",
"-e", "FIRESTORE_DATABASE_ID",
"-e", "GCP_PROJECT",
"-e", "GOOGLE_APPLICATION_CREDENTIALS=/app/gcp-credentials.json",
"-v", "${GOOGLE_APPLICATION_CREDENTIALS}:/app/gcp-credentials.json",
"ghcr.io/google-gemini/gemini-cli-issue-triage@sha256:e3de1523f6c83aabb3c54b76d08940a2bf42febcb789dd2da6f95169641f94d3"
],
"env": {
"GITHUB_TOKEN": "${GITHUB_TOKEN}",
"GEMINI_API_KEY": "${{ secrets.GEMINI_API_KEY }}",
"DATABASE_TYPE":"firestore",
"GCP_PROJECT": "${FIRESTORE_PROJECT}",
"FIRESTORE_DATABASE_ID": "(default)",
"GOOGLE_APPLICATION_CREDENTIALS": "${GOOGLE_APPLICATION_CREDENTIALS}"
},
"enabled": true,
"timeout": 600000
}
},
"maxSessionTurns": 25,
"coreTools": [
"run_shell_command(echo)",
"run_shell_command(gh issue comment)",
"run_shell_command(gh issue view)",
"run_shell_command(gh issue edit)"
],
"telemetry": {
"enabled": true,
"target": "gcp"
}
}
prompt: |-
## Role
You are an issue de-duplication assistant. Your goal is to find
duplicate issues, label the current issue as a duplicate, and notify
the user by commenting on the current issue, while avoiding
duplicate comments.
## Steps
1. **Find Potential Duplicates:**
- The repository is ${{ github.repository }} and the issue number is ${{ github.event.issue.number }}.
- Use the `duplicates` tool with the `repo` and `issue_number` to find potential duplicates for the current issue. Do not use the `threshold` parameter.
- If no duplicates are found, you are done.
- Print the JSON output from the `duplicates` tool to the logs.
2. **Refine Duplicates List (if necessary):**
- If the `duplicates` tool returns between 1 and 14 results, you must refine the list.
- For each potential duplicate issue, run `gh issue view <issue-number> --json title,body,comments` to fetch its content.
- Also fetch the content of the original issue: `gh issue view "${ISSUE_NUMBER}" --json title,body,comments`.
- Carefully analyze the content (title, body, comments) of the original issue and all potential duplicates.
- It is very important if the comments on either issue mention that they are not duplicates of each other, to treat them as not duplicates.
- Based on your analysis, create a final list containing only the issues you are highly confident are actual duplicates.
- If your final list is empty, you are done.
- Print to the logs if you omitted any potential duplicates based on your analysis.
- If the `duplicates` tool returned 15+ results, use the top 15 matches (based on descending similarity score value) to perform this step.
3. **Format Final Duplicates List:**
Format the final list of duplicates into a markdown string.
The format should be:
"Found possible duplicate issues:\n\n- #${issue_number}\n\nIf you believe this is not a duplicate, please remove the `status/possible-duplicate` label."
Add an HTML comment to the end for identification: `<!-- gemini-cli-deduplication -->`
4. **Check for Existing Comment:**
- Run `gh issue view "${ISSUE_NUMBER}" --json comments` to get all
comments on the issue.
- Look for a comment made by a bot (the author's login often ends in `[bot]`) that contains `<!-- gemini-cli-deduplication -->`.
- If you find such a comment, store its `id` and `body`.
5. **Decide Action:**
- **If an existing comment is found:**
- Compare the new list of duplicate issues with the list from the existing comment's body.
- If they are the same, do nothing.
- If they are different, edit the existing comment. Use
`gh issue comment "${ISSUE_NUMBER}" --edit-comment <comment-id> --body "..."`.
The new body should be the new list of duplicates, but with the header "Found possible duplicate issues (updated):".
- **If no existing comment is found:**
- Create a new comment with the list of duplicates.
- Use `gh issue comment "${ISSUE_NUMBER}" --body "..."`.
6. **Add Duplicate Label:**
- If you created or updated a comment in the previous step, add the `duplicate` label to the current issue.
- Use `gh issue edit "${ISSUE_NUMBER}" --add-label "status/possible-duplicate"`.
## Guidelines
- Only use the `duplicates` and `run_shell_command` tools.
- The `run_shell_command` tool can be used with `gh issue view`, `gh issue comment`, and `gh issue edit`.
- Do not download or read media files like images, videos, or links. The `--json` flag for `gh issue view` will prevent this.
- Do not modify the issue content or status.
- Only comment on and label the current issue.
- Reference all shell variables as "${VAR}" (with quotes and braces).

View File

@ -18,6 +18,7 @@ permissions:
id-token: 'write'
issues: 'write'
statuses: 'write'
packages: 'read'
jobs:
triage-issues:
@ -223,3 +224,105 @@ jobs:
- could also pertain to latency,
- other general software performance like, memory usage, CPU consumption, and algorithmic efficiency.
- Switching models from one to the other unexpectedly.
refresh-embeddings:
if: |-
${{ vars.TRIAGE_DEDUPLICATE_ISSUES != '' && github.repository == 'google-gemini/gemini-cli' }}
timeout-minutes: 20
runs-on: 'ubuntu-latest'
steps:
- name: 'Checkout repository'
uses: 'actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683'
- name: 'Generate GitHub App Token'
id: 'generate_token'
uses: 'actions/create-github-app-token@df432ceedc7162793a195dd1713ff69aefc7379e'
with:
app-id: '${{ secrets.APP_ID }}'
private-key: '${{ secrets.PRIVATE_KEY }}'
- name: Log in to GitHub Container Registry
uses: docker/login-action@v3
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: 'Run Gemini Issue Deduplication Refresh'
uses: 'google-github-actions/run-gemini-cli@20351b5ea2b4179431f1ae8918a246a0808f8747'
id: 'gemini_refresh_embeddings'
env:
GITHUB_TOKEN: '${{ steps.generate_token.outputs.token }}'
ISSUE_TITLE: '${{ github.event.issue.title }}'
ISSUE_BODY: '${{ github.event.issue.body }}'
ISSUE_NUMBER: '${{ github.event.issue.number }}'
REPOSITORY: '${{ github.repository }}'
FIRESTORE_PROJECT: '${{ vars.FIRESTORE_PROJECT }}'
with:
gcp_workload_identity_provider: '${{ vars.GCP_WIF_PROVIDER }}'
gcp_project_id: '${{ vars.GOOGLE_CLOUD_PROJECT }}'
gcp_location: '${{ vars.GOOGLE_CLOUD_LOCATION }}'
gcp_service_account: '${{ vars.SERVICE_ACCOUNT_EMAIL }}'
gemini_api_key: '${{ secrets.GEMINI_API_KEY }}'
use_vertex_ai: '${{ vars.GOOGLE_GENAI_USE_VERTEXAI }}'
use_gemini_code_assist: '${{ vars.GOOGLE_GENAI_USE_GCA }}'
settings: |-
{
"mcpServers": {
"issue_deduplication": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"--network", "host",
"-e", "GITHUB_TOKEN",
"-e", "GEMINI_API_KEY",
"-e", "DATABASE_TYPE",
"-e", "FIRESTORE_DATABASE_ID",
"-e", "GCP_PROJECT",
"-e", "GOOGLE_APPLICATION_CREDENTIALS=/app/gcp-credentials.json",
"-v", "${GOOGLE_APPLICATION_CREDENTIALS}:/app/gcp-credentials.json",
"ghcr.io/google-gemini/gemini-cli-issue-triage@sha256:e3de1523f6c83aabb3c54b76d08940a2bf42febcb789dd2da6f95169641f94d3"
],
"env": {
"GITHUB_TOKEN": "${GITHUB_TOKEN}",
"GEMINI_API_KEY": "${{ secrets.GEMINI_API_KEY }}",
"DATABASE_TYPE":"firestore",
"GCP_PROJECT": "${FIRESTORE_PROJECT}",
"FIRESTORE_DATABASE_ID": "(default)",
"GOOGLE_APPLICATION_CREDENTIALS": "${GOOGLE_APPLICATION_CREDENTIALS}"
},
"enabled": true,
"timeout": 600000
}
},
"maxSessionTurns": 25,
"coreTools": [
"run_shell_command(echo)"
],
"telemetry": {
"enabled": true,
"target": "gcp"
}
}
prompt: |-
## Role
You are a database maintenance assistant for a GitHub issue deduplication system.
## Goal
Your sole responsibility is to refresh the embeddings for all open issues in the repository to ensure the deduplication database is up-to-date.
## Steps
1. **Extract Repository Information:** The repository is ${{ github.repository }}.
2. **Refresh Embeddings:** Call the `refresh` tool with the correct `repo`. Do not use the `force` parameter.
3. **Log Output:** Print the JSON output from the `refresh` tool to the logs.
## Guidelines
- Only use the `refresh` tool.
- Do not attempt to find duplicates or modify any issues.
- Your only task is to call the `refresh` tool and log its output.