Skip to main content
Glama

🎯 What is CloudAgentSwarm?

CloudAgentSwarm is an autonomous incident response platform where specialized AI agents collaborate to triage, investigate, contain, and remediate cloud infrastructure outages.

Instead of a single monolithic prompt, tasks are distributed across role-specific agents that analyze telemetry, evaluate blast radius, enforce safety guardrails, and execute remediation with rollback guarantees.

 [Alert] ──► [Incident Commander] ──► [Telemetry Investigator]
                                               β”‚ (Root Cause)
                                               β–Ό
 [Safe Execution] ◄── [Guardrails & HITL] ◄── [Blast Radius Analyst]
        β”‚                                      β”‚
        β–Ό                                      β–Ό
 [Post-Mortem] ◄────────────────────── [Remediation Planner]

Related MCP server: devops-mcp-server

πŸ€– Specialized Agents

Agent

Role

Focus

Incident Commander

War Room Lead

Severity classification (SEV-1 to SEV-4) and consensus orchestration.

Telemetry Investigator

Observability

Correlates CloudWatch metrics, PromQL queries, and pod logs to find root causes.

Blast Radius Analyst

Safety & Security

Evaluates IAM privilege risks, VPC boundaries, and financial exposure.

Remediation Planner

Site Reliability

Generates atomic remediation actions with deterministic rollback commands.


πŸ›‘οΈ Safety & Guardrails

  • Hard Guardrails: Destructive operations (drop_database, delete_vpc, destroy_kms_key) are blocked at policy level.

  • Human-in-the-Loop (HITL): Actions with risk score $> 40.0$ or flagged as destructive mandate interactive human approval.

  • Rollback Guarantee: Every action requires a verified inverse rollback command before execution.


πŸš€ Quick Start

1. Installation

git clone https://github.com/ekrmcakir/cloud-agent-swarm.git
cd cloud-agent-swarm
pip install -e .

2. Run Chaos Simulations

# List available scenarios
cloud-swarm list-scenarios

# Run Kubernetes OOM cascade simulation
cloud-swarm simulate --scenario oom_cascade

# Run with JSON output
cloud-swarm simulate --scenario iam_key_leak --json

πŸ”Œ Model Context Protocol (MCP)

CloudAgentSwarm runs as a native MCP stdio server, allowing Claude Desktop, Cursor, or Antigravity to trigger investigations:

{
  "mcpServers": {
    "cloud-agent-swarm": {
      "command": "cloud-swarm",
      "args": ["mcp"]
    }
  }
}

Exposed Tools:

  • list_scenarios β€” Lists runnable chaos outage scenarios.

  • trigger_investigation β€” Runs the swarm against an incident.

  • get_postmortem β€” Fetches the generated SRE post-mortem report.


πŸ”₯ Included Chaos Scenarios

  1. oom_cascade (SEV-1): Kubernetes pod memory exhaustion (Exit code 137) triggering 502/504 errors.

  2. iam_key_leak (SEV-1): Leaked administrator IAM key with unauthorized S3 data scraping.

  3. dynamodb_throttling (SEV-2): Capacity burst exceeding provisioned RCU on hot partition.

  4. split_brain_partition (SEV-2): Transit Gateway route drop causing cross-AZ communication failure.


πŸ§ͺ Testing

make test
# or
python3 -m unittest discover -s tests -p "test_*.py"

πŸ“œ License

Apache 2.0 Β© 2026 Ekrem Cakir.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables autonomous SRE incident investigation by allowing users to describe incidents in natural language. The agent follows a governed state machine to gather read-only evidence and produce grounded conclusions.
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Exposes DevOps/SRE operations like Kubernetes pod status, Prometheus metrics queries, and deploy history as tools for MCP-compatible clients.
      -
    • F
      license
      Not graded
      quality
      D
      maintenance
      Provides MCP tools to search engineering runbooks and historical incidents using semantic retrieval, supporting evidence-grounded incident investigation.
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI agents to investigate production incidents by exposing service health, logs, and deployment data through MCP tools.
      9 npm
      ISC