Your monitoring stack can tell you the pod is dying. 01Agents brings it back.
An open-source Kubernetes incident resolution engine. It unifies real-time cluster triage (L1), secondary telemetry reconstruction (L2), and deep kernel eBPF & network packet intelligence (L3) via a 100% pluggable, unlimited-extensibility plugin architecture—diagnosing root causes from container state down to raw network packets and kernel scheduler throttling.
Pluggable Observability & GitOps Stack
Zero application code modifications required. 01Agents plug dynamically into your cluster's existing telemetry collectors, log streams, metric stores, GitOps engines, and kernel eBPF probes via the Plugin Marketplace.
SigNoz
Native integration with SigNoz v5 for historical ClickHouse log queries & distributed APM tracing.
Grafana Loki
Direct Loki Tail WS proxy & PVC volume mounting to retrieve logs even from terminated & deleted pods.
Grafana Alloy
Next-gen OpenTelemetry pipeline & agent for streaming node-level metrics and container logs.
Prometheus
PromQL metric parsing for instant CPU, memory, OOMKilled, and crash-loop backoff detection.
ELK Stack
Seamless search & indexed log analysis via Elasticsearch REST APIs and Kibana dashboard audit sync.
Datadog
Bi-directional Datadog monitor webhook intake & automated incident remediation triggers.
ArgoCD
Automated GitOps state verification, commit rollback, and sync status checks upon incident resolution.
FluxCD
Native reconciliation with Flux Kustomization and HelmRelease objects for safe cluster state restoration.
Why Standard K8s Diagnostic Tools Fail on Pod Termination
When a pod enters CrashLoopBackOff or gets evicted due to OOMKilled, Kubernetes eventually removes the pod resource. Running kubectl logs or querying the API server at that point returns:
01Agents solves this with a 3-tier multi-agent triad—each tier escalating to the next when standard diagnostics hit their limits:
SKILL.md runbooks—no hardcoded failure codes.Domain-Specific Intelligence Across L1 Triage, L2 Secondary Telemetry & L3 eBPF Kernel Wire Analysis
Not limited to fixed failure codes or static agents. 01Agents features a dynamic SKILL.md runbook engine and a pluggable subagent registry—enabling teams to introduce unlimited custom skills, GitOps workflows, and kernel diagnostic probes without rebuilding Docker images or restarting pods.
TCP Socket & Packet Flow Inspector
Traces TCP state transitions, connection resets (RST), round-trip time (RTT), and silent packet drops at the Linux network stack via eBPF kprobes and XDP hooks.
CoreDNS & Wire-Protocol Analyzer
Sniffs raw socket traffic on UDP/TCP port 53 and HTTP/gRPC headers to catch NXDOMAIN cascades, DNS query timeouts, and microservice inter-pod delays—without any proxy.
Kernel CFS Scheduler & Resource Contention
Detects Linux Completely Fair Scheduler (CFS) CPU quota throttling, memory allocation latency, and dirty page flushes invisible to Prometheus 30-second scrape intervals.
Ghost Incident & Invisible Failure Correlator
Cross-correlates kernel-space anomalies with application APM traces to diagnose incidents where pods appear healthy but transactions fail silently with no container crash.
Pluggable Telemetry Reconstruction
Plugs seamlessly into your telemetry plugins (SigNoz, Grafana Loki, Prometheus, ELK, Datadog, and any custom backend) to pull pre-failure logs, metric trends, and APM traces after pod deletion.
Multi-Agent Root Cause Analysis
Executes a 4-agent LangGraph workflow to cross-correlate log stacktraces, OOM events, Git commit history, and eBPF wire evidence via installed plugins.
Human-in-the-Loop Safeguards
Auto-applies low-risk remedies while queuing high-risk cluster state mutations for SRE confirmation via Slack or API plugins.
CrashLoop Triage
Diagnoses container crash loops, missing env variables, panics, and unhandled exception stack traces.
Memory Exhaustion
Traces heap spikes, memory leaks, container memory limit breaches, and recommends scaled limits.
Registry & Image Audit
Verifies container image availability, registry authentication tokens, and secret volume mounts.
Runtime Diagnostics
Identifies container runtime failures, cgroup misconfigurations, and volume mount permission errors.
Scheduling Taints & Affinities
Analyzes node taint mismatches, unschedulable pod requirements, and node resource shortfalls.
Process Exit Analysis
Correlates process exit codes (e.g. exit 137, 127) with binary startup scripts and entrypoint parameters.
Node Resource Pressure
Detects disk, memory, and PID pressure conditions triggering pod eviction and node cordoning.
Eviction Root Cause
Traces eviction history, priority class conflicts, resource quota exhaustion, and disruption budgets.
Bring Your Own SKILL.md
Define any failure scenario as a declarative SKILL.md runbook. 01Agents hot-loads it instantly—no rebuild, no pod restart required.
SKILL.md runbook with YAML triggers, required tools, and remediation steps. 01Agents hot-loads it at runtime via the dynamic SkillRegistry—no Docker rebuild, no pod restart, no core code changes.Pluggable Telemetry Ecosystem
Services like SigNoz, Prometheus, ELK, Datadog and Grafana Loki, ArgoCD, Cilium Hubble, and Pixie are not hardcoded. They are decoupled plugins that users can install, uninstall, or replace with their own custom plugins via the Plugin Marketplace. No limit on the number of plugins.
01Agents Plugin Marketplace
Browse, install, and configure plugins for L1 triage, L2 telemetry, L3 eBPF, and custom GitOps backends.
SigNoz Telemetry Plugin
Extract historical ClickHouse traces & APM metrics pre-and-post pod eviction.
Prometheus Metrics Plugin
PromQL metric parsing & threshold analysis for CPU, RAM, and OOMKilled events.
ELK Stack Plugin
Seamless Elasticsearch query engine and Kibana dashboard audit synchronizer.
Datadog Integration Plugin
Webhook intake and incident context enrichment from Datadog monitors.
Grafana Loki WS Plugin
Loki WebSocket Tail proxy for persistent live log ingestion and PVC snapshotting.
ArgoCD GitOps Plugin
Automated GitOps state verification, commit rollback, and sync status checks upon incident resolution.
FluxCD Reconciliation Plugin
Native reconciliation with Flux Kustomization and HelmRelease objects for safe cluster state restoration.
Cilium Hubble PluginL3
Query Hubble flow metrics, packet drop reasons, and per-pod network policies via gRPC for L3 wire-level diagnostics.
Pixie Auto-Tracing PluginL3
Auto-traces HTTP, gRPC, PostgreSQL, and MySQL inter-pod protocols at the socket level without sidecars or code changes.
VictoriaMetrics Plugin
High-performance time-series queries via VictoriaMetrics MetricsQL for large-scale metric reconstruction.
OpenSearch Log Plugin
Full-text log search and aggregation via OpenSearch REST APIs for postmortem log reconstruction.
Bring Your Own PluginBYO
Build & plug your proprietary internal observability, ticketing, or ITSM tool using the Python/TypeScript Plugin SDK.
Zero Lock-In Architecture
Switch observability backends anytime without changing agent core code.
BYO Custom Plugin SDK
Write custom telemetry handlers in Python or TypeScript with a standard interface.
One-Click Hot-Swapping
Enable or disable plugins live without restarting 01Agent pods.
eBPF Kernel Plugins
Cilium Hubble, Pixie, and Inspektor Gadget plugins for L3 wire-level diagnostics.
Enterprise-Grade Observability & Instant Incident Auditability
Every L1 triage decision, L2 secondary telemetry analysis, and L3 eBPF kernel wire finding is fully logged, exportable, and queryable. 01Agents integrate natively into your existing observability and GitOps stack without modifying application code.
Decision history is queryable via API, exportable in JSON or CSV, and structured for postmortem review. When something needs to be explained — the exact evidence, eBPF wire telemetry, and RCA path are ready for audit.
Engineered for Production Kubernetes Clusters
Designed from the ground up to solve postmortem log loss, surface invisible kernel-level incidents, prevent accidental cluster mutations, and integrate natively into existing DevOps tooling—with unlimited extensibility for custom skills and backends.
Postmortem Log Reconstruction
Queries ClickHouse (SigNoz) & Loki WebSocket streams to retrieve logs after pods are deleted or terminated.
Deterministic HITL Approval Gates
Classifies remediation actions by risk level. High-risk mutations (scaling, config patches) require operator approval.
LangGraph 4-Agent Workflow
Orchestrates dedicated agents for Triage, Telemetry Querying, Root Cause Analysis (RCA), and Remediation Generation.
GitOps Rollback Verification
Verifies application declarative state with ArgoCD and FluxCD before marking any incident resolved.
Zero Application Code Instrumentation
Operates entirely at the cluster control plane & telemetry backend layer without requiring app SDK changes.
Exportable Incident Decision Records
Outputs structured JSON/CSV decision traces for postmortems, compliance reviews, and team handoffs.
Sub-Millisecond Wire-Level Telemetry
Level-3 eBPF probes trace TCP socket state machines, DNS query latency, packet drops, and CFS scheduler throttling invisible to standard K8s observability.
Infinite SKILL.md Hot-Loading
Custom failure remediation skills are introduced as declarative SKILL.md runbooks and loaded just-in-time by the SkillRegistry—zero rebuild, zero pod restart.
Five Steps from Incident to Verified Fix
How 01Agents integrates into your deployment pipeline and handles pod failures from detection through L3 kernel wire analysis to verified GitOps remediation.
Cluster Event Monitoring & Ingestion
Deploys non-invasively into your Kubernetes cluster to listen for real-time pod events, OOMKilled signals, and state transitions via native Kafka event streams. L1 triage skills are fully customizable via SKILL.md runbooks—not limited to any fixed set of failure codes.
Pluggable Telemetry Backend Query
When a pod terminates and returns 404, 01Agents automatically initiates a lookback query across installed telemetry plugins (SigNoz, Grafana Loki, Prometheus, ELK Stack, Datadog, or any custom backend). The plugin architecture is 100% open—you can introduce unlimited custom observability backends via plugin.json and the MCP standard.
L3 eBPF Kernel Escalation (When Logs Are Inconclusive)
NEW · L3 eBPFIf application logs, APM traces, and K8s events don't fully explain the incident, 01Agents escalates to Level-3. The L3 eBPF agent deploys in-kernel probes to trace raw TCP packet drops, DNS query latency, CoreDNS NXDOMAIN cascades, Linux CFS scheduler throttling, and socket buffer overflows—surfacing invisible failures that standard observability completely misses.
Multi-Agent Root Cause Analysis
The L2 LangGraph engine cross-correlates pre-failure logs, metric memory spikes, container configuration history, and L3 eBPF wire evidence to pinpoint the exact root cause with a measurable RCA confidence score.
Human-in-the-Loop Safeguards & GitOps Sync
Low-risk remedies execute automatically, while high-risk cluster changes pause for SRE confirmation before verifying manifest restoration with ArgoCD or FluxCD. Every decision is logged with full audit provenance, exportable as JSON or CSV.
Start Resolving Kubernetes Incidents Automatically
Install the 01Agents operator on your cluster, enable your preferred telemetry plugins (SigNoz, Loki, Prometheus, ELK), and stop losing logs when pods are deleted.