Land With Full Context
Start from a Flashduty incident or IM war room. Flashduty AI SRE automatically pulls alert, service, team, and on-call context.
The AI SRE that knows your stack
Flashduty AI SRE shortens the path from alert to resolution — automating investigation, root cause analysis, and next-step recommendations while you stay in control.
API rate limit investigation / Flashcat AI SRE
Context15%·2m 43s
Add a follow-up... Press Enter to queue
Type / for commandsflashduty-05No team selected
AutoPersonal
AI generated, for reference only
1.0SCENARIOS
Cover the Full Incident Lifecycle From Trigger to Review
From alert intake and automated diagnosis to post-incident review, Flashduty AI SRE preserves service, change, ownership, and remediation context across the workflow.
AI SRE WORKBENCH
INC-482913 order-api P1 latency
Order service P1 incident. Latency just spiked to 8s. Find the root cause and sync the conclusion to the war room.
Plan: pull recent changes, related alerts, and dependency paths,run 3 steps in parallel…
Ran query_metrics (service="order-api", window="30m")
64msRan list_recent_changes (service="order-api", since="2h")
92msRan fetch_runbook (ref="@duty/order/p1-latency.md")
48msPreliminary Finding
Likely cause
DB pool
Trigger change
#4821
Affected nodes
12 / 36
Posted to war room · #war-room-482913 · notified @oncall @li.wei @yang.bai @hang.yu
Proactive Console Triage
Ask why payment-api latency is rising. Flashduty AI SRE checks alerts, logs, changes, and dependency health, then returns likely causes and next actions.
War-Room Auto-Diagnose
As soon as the incident room is created, Flashduty AI SRE runs the first investigation and posts early findings back so the team starts from evidence.
Summon From IM
Mention Flashduty AI SRE in a group or private chat without changing tools. Everyone sees the same investigation thread, findings, and evidence trail.
Operational Review Insights
Use /insight to review the last 30 days of Flashduty AI SRE sessions, spot repeat issues, missing runbooks, and copied context that should become system improvements.
2.0HOW IT WORKS
From the First Alert, AI SRE Owns the First Investigation
When an alert enters Flashduty, Flashduty AI SRE correlates related signals, plans the investigation path, and syncs verifiable findings to the response team.
Plan Its Own Path
Engineers do not have to script each step. Describe the problem and the Agent decomposes work, selects tools, and streams its investigation.
Call Tools and Run Checks
Use Skills, MCP, external Agents, or self-hosted Runners to access monitors, logs, files, command lines, and internal systems.
Ship Evidence-Backed Findings
The output is not a guess. It includes the investigation path, intermediate findings, and next-step recommendations ready for the war room.
Compound Into Knowledge
Common diagnostic flows become Skills, long-lived context becomes Knowledge Packs, and the next response reuses what the team learned.
3.0CAPABILITIES
Not Just Chat, an SRE That Can Do the Work
Flashduty AI SRE goes beyond answers: it uses Skills, MCP, and Knowledge Packs to run checks, call tools, assemble evidence, and produce traceable triage conclusions.
Conversational Triage
Describe an anomaly, alert, change, or service problem in natural language. Flashduty AI SRE streams its plan, tool calls, intermediate findings, and final conclusion.
IM War-Room Linkage
Mention Flashduty AI SRE in Feishu/Lark, DingTalk, WeCom, or Slack to start or continue triage. When the room opens, it can run an initial diagnosis and post findings.
Skill Triage Flows
Package common runbooks, inspection steps, and diagnostic procedures as Skills. Once enabled, the Agent loads them on demand in the right scenario.
Knowledge Pack Memory
Use service inventories, on-call paths, triage history, and long-lived context so Flashduty AI SRE understands not just one incident, but your system.
MCP External Tool Access
Connect external data sources and tools through Model Context Protocol. The Agent connects, executes, and disconnects as needed.
BYOC Self-Hosted Runtime
Use the Flashduty cloud sandbox by default, or deploy a self-hosted Runner so Flashduty AI SRE can execute diagnostic tasks inside your network.
4.0INTEGRATIONS
Connect Your Existing Stack, No Migration Required
Keep your existing IM, incident response, observability, and runtime tools, and bring diagnostic automation into the workflows your team already uses.
IM Collaboration
SlackFeishu LarkDingTalkWeCom
Incident Response
FlashdutyPagerDutyOpsgenieWar-room
Observability
PrometheusGrafanaOpenTelemetryLoki
Runtime
BYOC RunnerCloud SandboxMCP ToolsA2A Agents
5.0SECURITY & RUNTIME
Reach Internal Systems Without Losing Execution Boundaries
For internal diagnostics and controlled execution, use hosted sandboxes or self-hosted Runners while preserving access scope, permissions, and runtime isolation.
cloud-sandbox / runner shell
isolated$fd-sandbox run ai-sre --task incident-4521 --scope read-only
inithosted sandbox ready, ephemeral workspace 15m TTL
runnerinternal diagnostics routed through byoc-runner-vpc-prod
mcpprometheus / loki mounted on demand, disconnected after task
policyKnowledge and plugins load by account / team role
Hosted Sandbox to Start Fast
Self-Hosted Runner for Internal Networks
MCP Connects on Demand, Not by Default
Knowledge Scoped by Account and Team
Roles Gate Visibility and Capabilities
Flashduty AI SRE supports both hosted cloud sandboxes and self-hosted Runners. Teams that need access to internal services, private logs, or proprietary tools can deploy BYOC Runners so diagnostics happen inside a controlled environment. Menus, knowledge, plugins, and runtime visibility are governed by account roles.
Start quickly in the hosted sandbox, or deploy a self-hosted Runner inside your VPC. Menus, knowledge, plugins, and runtime visibility are all gated by account roles.
Frequently Asked Questions
Less noise. Faster recovery.
Start free—no credit card. Book a demo to see your alerting pipeline end to end.