Skip to main content
Glama
itsme825

github-incident-mcp

by itsme825
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Agentic Incident Response Pipeline

An autonomous, multi-agent incident response pipeline built with LangGraph, LangChain, and official FastMCP (Model Context Protocol) servers for real-time incident triage, root-cause diagnostics, automated remediation, human sign-off state persistence, and Slack notification dispatching.

🌟Architecture Overview

sequenceDiagram
    autonumber
    actor Alert as  Alert Trigger (PagerDuty / GitHub / Prometheus)
    participant Triage as  Triage Agent
    participant OSV as  OSV MCP Server
    participant Diag as  Diagnostic Agent
    participant GH as  GitHub MCP Server
    participant Remed as  Remediation Agent
    actor Gate as  Human Sign-off Gate
    participant Exec as  Execution Agent
    participant Slack as  Slack Channel

    Alert->>Triage: Ingest Alert Payload
    Triage->>OSV: Tool Call: osv_scan_alert_payload()
    OSV-->>Triage: Vulnerability Advisories & Severity
    Triage->>Diag: TriageResult (Sev1-Sev4)
    Diag->>GH: Tool Call: github_list_recent_prs()
    GH-->>Diag: Merged PRs & Diff Summaries
    Diag->>Diag: Temporal Correlation (Error spike vs. PR merge)
    Diag->>Remed: DiagnosticResult & Root Cause
    Remed->>Gate: Draft Proposal (Pauses at Sign-off Gate)
    
    alt Approve
        Gate->>Exec: Approve Command
        Exec->>GH: Tool Call: github_execute_revert_pr()
        Exec->>Slack: Dispatch Incident Resolution Card
    else Edit Command / Reject / Escalate
        Gate->>Slack: Execute Overridden Command or Escalate
    end

Related MCP server: GitHub MCP Server

Key Features

  1. Triage Agent & Security Cross-Check:

    • Classifies incident severity (Sev1 to Sev4) and affected component.

    • Integrates with OSV MCP Server (https://api.osv.dev/v1/query) to cross-check dependency vulnerabilities (CVEs).

  2. Diagnostic Agent & Temporal Root Cause Correlation:

    • Interfaces with GitHub FastMCP Server to query recently merged PRs and commit diffs.

    • Correlates error log spikes and metric anomalies with PR merge timestamps.

  3. Remediation Agent & Feedback Loop:

    • Proposes targeted remediation actions (Git PR reverts or deployment rollbacks).

    • Re-plans proposals based on human engineer feedback if rejected at the gate.

  4. Human Sign-off Gate (LangGraph State Graph Persistence):

    • Genuinely pauses graph execution using interrupt() and MemorySaver.

    • Supports 4 interactive actions: APPROVE, EDIT_COMMAND, REJECT, ESCALATE.

  5. Execution Agent & Slack Notifications:

    • Executes approved PR reverts programmatically via GitHub REST API.

    • Dispatches rich Slack cards to #oncall-incidents or escalation alerts to #senior-oncall-team.

Repository Structure

├── src/
│   ├── agents/            # Triage, Diagnostic, Remediation, Execution Agents
│   ├── mcp/               # GitHub & OSV FastMCP Servers (stdio transport)
│   ├── models/            # Pydantic schemas (Alert, Triage, Diagnostic, Remediation, Execution)
│   ├── pipeline/          # LangGraph IncidentPipeline state graph
│   ├── services/          # GitHub, OSV, Log/Metrics, Notification Services
│   ├── utils/             # Logging and utility helpers
│   └── config.py          # Centralized configuration settings
├── tests/
│   ├── test_mcp_servers.py
│   ├── test_pipeline_graph.py
│   └── run_test_scenario.py # E2E Multi-Agent Scenario Suite
├── .env.example
├── .gitignore
├── requirements.txt
└── README.md

Installation & Setup

  1. Clone Repository & Install Dependencies:

    git clone https://github.com/<YOUR_USERNAME>/agentic-debugging-project.git
cd agentic-debugging-project
pip install -r requirements.txt

  2. Configure Environment Variables: Copy .env.example to .env and set your credentials:

    cp .env.example .env

    Edit .env:

    GEMINI_API_KEY=your_gemini_api_key_here
GITHUB_TOKEN=your_github_personal_access_token_here
LLM_PROVIDER=gemini
LLM_MODEL=gemini-2.5-flash

Testing & Verification

  1. Run Unit Tests:

    python -m pytest tests/

  2. Run E2E Incident Scenarios:

    python -m tests.run_test_scenario

  3. Run Stdio FastMCP Servers:

    • GitHub FastMCP Server:

      python src/mcp/github_mcp_server.py

    • OSV FastMCP Server:

      python src/mcp/osv_mcp_server.py

Claude Desktop MCP Integration

Add the following to your %APPDATA%\Claude\claude_desktop_config.json:

{
  "mcpServers": {
    "github-incident-mcp": {
      "command": "python",
      "args": [
        "C:\\path\\to\\agentic_debugging_project_github\\src\\mcp\\github_mcp_server.py"
      ],
      "env": {
        "GITHUB_TOKEN": "your_github_personal_access_token_here",
        "PYTHONUNBUFFERED": "1"
      }
    },
    "osv-vulnerability-mcp": {
      "command": "python",
      "args": [
        "C:\\path\\to\\agentic_debugging_project_github\\src\\mcp\\osv_mcp_server.py"
      ],
      "env": {
        "OSV_API_URL": "https://api.osv.dev/v1/query",
        "PYTHONUNBUFFERED": "1"
      }
    }
  }
}

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers