Feat/Issue Monitoring Agent

.github/workflows/issue-monitor.yml

@@ -0,0 +1,61 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: ADK Issue Monitoring Agent
+
+on:
+  schedule:
+    # Runs daily at 6:00 AM UTC
+    - cron: '0 6 * * *'
+
+  # Allows manual triggering from the GitHub Actions tab
+  workflow_dispatch:
+    inputs:
+      full_scan:
+        description: 'Run an Initial Full Scan of ALL open issues'
+        required: false
+        type: boolean
+        default: false
+
+jobs:
+  sweep-spam:
+    runs-on: ubuntu-latest
+    timeout-minutes: 120 
+    permissions:
+      issues: write
+      contents: read
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+
+      - name: Set up Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: '3.11'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install requests google-adk python-dotenv
+
+      - name: Run Issue Monitoring Agent
+        env:
+          GITHUB_TOKEN: ${{ secrets.ADK_TRIAGE_AGENT }}
+          GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
+          OWNER: ${{ github.repository_owner }}
+          REPO: ${{ github.event.repository.name }}
+          CONCURRENCY_LIMIT: 3
+          INITIAL_FULL_SCAN: ${{ github.event.inputs.full_scan == 'true' }}
+        run: python -m spam_sweeper_agent.main

contributing/samples/adk_issue_monitoring_agent/PROMPT_INSTRUCTION.txt

@@ -0,0 +1,18 @@
+You are the automated security and moderation agent for the {OWNER}/{REPO} repository.
+
+You will be provided with an Issue Number and a list of comments made by non-maintainers.
+Your job is to read through these comments and identify if any of them contain SPAM, promotional content for 3rd-party websites, SEO links, or objectionable material.
+
+CRITERIA FOR SPAM:
+- The comment is completely unrelated to the repository or the specific issue.
+- The comment promotes a 3rd party product, service, or website.
+- The comment is generic "SEO spam" (e.g., "Great post! Check out my site at [link]").
+
+INSTRUCTIONS:
+1. Evaluate the provided comments.
+2. If you identify spam, call the `flag_issue_as_spam` tool. 
+   - Pass the `item_number`.
+   - Pass a brief `detection_reason` explaining which comment is spam and why (e.g., "@spammer_bot posted an irrelevant link to a shoe store").
+3. If NONE of the comments contain spam, do NOT call any tools. Just respond with "No spam detected."
+
+Remember: Do not flag comments that are merely unhelpful, off-topic, or from beginners asking legitimate questions. Only flag actual spam, endorsements, or objectionable material.

contributing/samples/adk_issue_monitoring_agent/README.md

@@ -0,0 +1,62 @@
+# ADK Issue Monitoring Agent 🛡️
+
+An intelligent, cost-optimized, automated moderation agent built with the **Google Agent Development Kit (ADK)**. 
+
+This agent automatically audits GitHub repository issues to detect SEO spam, unsolicited promotional links, and irrelevant third-party endorsements. If spam is detected, it automatically applies a `spam` label and alerts the repository maintainers.
+
+## ✨ Key Features & Optimizations
+
+* **Zero-Waste LLM Invocations:** Fetches issue comments via REST APIs and pre-filters them in Python. It automatically ignores comments from maintainers, `[bot]` accounts, and the official `adk-bot`. The Gemini LLM is never invoked for safe threads, saving 100% of the token cost.
+* **Dual-Mode Scanning:** Can perform a **Deep Clean** (auditing the entire history of all open issues) or a **Daily Sweep** (only fetching issues updated within the last 24 hours).
+* **Token Truncation:** Uses Regular Expressions to strip out Markdown code blocks (` ``` `) replacing them with `[CODE BLOCK REMOVED]`, and truncates unusually long text to 1,500 characters before sending it to the AI.
+* **Idempotency (Anti-Double-Posting):** The bot reads the comment history for its own signature. If it has already flagged an issue, it instantly skips it, preventing infinite feedback loops.
+
+---
+
+## Configuration
+
+The agent is configured via environment variables, typically set as secrets in GitHub Actions.
+
+### Required Secrets
+
+| Secret Name | Description |
+| :--- | :--- |
+| `GITHUB_TOKEN` | A GitHub Personal Access Token (PAT) or Service Account Token with `repo` and `issues: write` scope. |
+| `GOOGLE_API_KEY` | An API key for the Google AI (Gemini) model used for reasoning. |
+
+### Optional Configuration
+
+These variables control the scanning behavior, thresholds, and model selection.
+
+| Variable Name | Description | Default |
+| :--- | :--- | :--- |
+| `INITIAL_FULL_SCAN` | If `true`, audits every open issue in the repository. If `false`, only audits issues updated in the last 24 hours. | `false` |
+| `SPAM_LABEL_NAME` | The exact text of the label applied to flagged issues. | `spam` |
+| `BOT_NAME` | The GitHub username of your official bot to ensure its comments are ignored. | `adk-bot` |
+| `CONCURRENCY_LIMIT` | The number of issues to process concurrently. | `3` |
+| `SLEEP_BETWEEN_CHUNKS` | Time in seconds to sleep between batches to respect GitHub API rate limits. | `1.5` |
+| `LLM_MODEL_NAME`| The specific Gemini model version to use. | `gemini-2.5-flash` |
+| `OWNER` | Repository owner (auto-detected in Actions). | (Environment dependent) |
+| `REPO` | Repository name (auto-detected in Actions). | (Environment dependent) |
+
+---
+
+## Deployment
+
+To deploy this agent, a GitHub Actions workflow file (`.github/workflows/issue-monitor.yml`) is recommended.
+
+### Directory Structure Note
+Because this agent resides within the `adk-python` package structure, the workflow must ensure the script is executed correctly to handle imports. It must be run as a module from the parent directory.
+
+### Example Workflow Execution
+```yaml
+      - name: Run ADK Issue Monitoring Agent
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
+          OWNER: ${{ github.repository_owner }}
+          REPO: ${{ github.event.repository.name }}
+          # Mapped to the manual trigger checkbox in the GitHub UI
+          INITIAL_FULL_SCAN: ${{ github.event.inputs.full_scan == 'true' }}
+          PYTHONPATH: contributing/samples
+        run: python -m adk_issue_monitoring_agent.main

contributing/samples/adk_issue_monitoring_agent/__init__.py

@@ -0,0 +1,15 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from . import agent

contributing/samples/adk_issue_monitoring_agent/agent.py

@@ -0,0 +1,87 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+import os
+from typing import Any
+
+from adk_issue_monitoring_agent.settings import GITHUB_BASE_URL
+from adk_issue_monitoring_agent.settings import LLM_MODEL_NAME
+from adk_issue_monitoring_agent.settings import OWNER
+from adk_issue_monitoring_agent.settings import REPO
+from adk_issue_monitoring_agent.settings import SPAM_LABEL_NAME
+from adk_issue_monitoring_agent.utils import error_response
+from adk_issue_monitoring_agent.utils import post_request
+from google.adk.agents.llm_agent import Agent
+from requests.exceptions import RequestException
+
+logger = logging.getLogger("google_adk." + __name__)
+
+
+def load_prompt_template(filename: str) -> str:
+  file_path = os.path.join(os.path.dirname(__file__), filename)
+  with open(file_path, "r") as f:
+    return f.read()
+
+
+PROMPT_TEMPLATE = load_prompt_template("PROMPT_INSTRUCTION.txt")
+
+# --- Tools ---
+
+
+def flag_issue_as_spam(
+    item_number: int, detection_reason: str
+) -> dict[str, Any]:
+  """
+  Flags an issue as spam by adding a label and leaving a comment for maintainers.
+
+  Args:
+      item_number (int): The GitHub issue number.
+      detection_reason (str): The explanation of what the spam is.
+  """
+  logger.info(f"Flagging #{item_number} as SPAM. Reason: {detection_reason}")
+
+  label_url = (
+      f"{GITHUB_BASE_URL}/repos/{OWNER}/{REPO}/issues/{item_number}/labels"
+  )
+  comment_url = (
+      f"{GITHUB_BASE_URL}/repos/{OWNER}/{REPO}/issues/{item_number}/comments"
+  )
+
+  alert_body = (
+      "🚨 **Automated Spam Detection Alert** 🚨\n"
+      "@maintainers, a suspected spam comment was detected in this thread.\n\n"
+      f"**Reason:** {detection_reason}"
+  )
+
+  try:
+    # 1. Add Label
+    post_request(label_url, {"labels": [SPAM_LABEL_NAME]})
+    # 2. Post Alert Comment
+    post_request(comment_url, {"body": alert_body})
+    return {"status": "success", "message": "Maintainers alerted successfully."}
+  except RequestException as e:
+    return error_response(f"Error flagging issue: {e}")
+
+
+root_agent = Agent(
+    model=LLM_MODEL_NAME,
+    name="spam_auditor_agent",
+    description="Audits issue comments for spam.",
+    instruction=PROMPT_TEMPLATE.format(
+        OWNER=OWNER,
+        REPO=REPO,
+    ),
+    tools=[flag_issue_as_spam],
+)