Omni Resilience
Monitor provider health, circuit-breaker states, p50/p95/p99 latency metrics, and budget guard alerts. Inspect connection cooldowns and model lockouts in real time.
Get the complete skill folder, including its scripts and reference files.
When to use this skill
Use Omni Resilience when an AI agent needs a reusable SKILL.md workflow for this job: Monitor provider health, circuit-breaker states, p50/p95/p99 latency metrics, and budget guard alerts. Inspect connection cooldowns and model lockouts in real time.
When not to use it
Skip Omni Resilience when the task is outside the analysis category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.
How to install
Where will you use it?
mkdir -p ~/.claude/skills/omni-resilience # Save SKILL.md to ~/.claude/skills/omni-resilience/SKILL.md
- Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/omni-resilience/SKILL.md. For a shared project, commit the folder at .claude/skills/omni-resilience/ instead.
- Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
- For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.
Full install guide for Claude, Cursor, and Codex →
Try a starter prompt
After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.
Use the Omni Resilience skill for this task: Use Omni Resilience when this documented workflow matches the task. Confirm any required inputs or files before making changes.
What supports this recommendation?
71/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.
- Documentation
- 77/100
- Practical detail
- 68/100
- Source evidence
- 63/100
- Source signals
- 72/100
Model compatibility evidence
- claudeinferred
The skill text mentions Claude or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
- chatgptuntested
No model-specific signal or recorded compatibility test was found.
- geminiuntested
No model-specific signal or recorded compatibility test was found.
- copilotuntested
No model-specific signal or recorded compatibility test was found.
- llamauntested
No model-specific signal or recorded compatibility test was found.
- perplexityuntested
No model-specific signal or recorded compatibility test was found.
- mistraluntested
No model-specific signal or recorded compatibility test was found.
- grokuntested
No model-specific signal or recorded compatibility test was found.
Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.
What this skill does
<!-- generated by src/lib/agentSkills/generator.ts; manual edits will be overwritten -->
## Overview
Monitor provider health, circuit-breaker states, p50/p95/p99 latency metrics, and budget guard alerts. Inspect connection cooldowns and model lockouts in real time.
## Authentication
All requests require a valid Bearer token or session cookie. Obtain a token via `POST /api/auth/login` or configure `REQUIRE_API_KEY=false` for local development.
## Endpoints
### GET /api/monitoring/health
System health check
Returns system health including uptime, memory, circuit breakers, rate limits
```bash curl http://localhost:20128/api/monitoring/health \ -H "Authorization: Bearer $OMNIROUTE_TOKEN" ```
### GET /api/monitoring/model-health
Get per-model health states
Management-only report that classifies every synced model on every provider connection as HEALTHY, DEGRADED, RATE_LIMITED, AUTH_FAILED, PROVIDER_DOWN, CONNECTION_BLOCKED, MODEL_LOCKED or UNKNOWN, from the connection state, the provider circuit breaker and the per-model lockout. Includes connection ids and names, so it requires a management principal.
```bash curl http://localhost:20128/api/monitoring/model-health \ -H "Authorization: Bearer $OMNIROUTE_TOKEN" ```
### GET /api/monitoring/compression
Get compression result-memo statistics
In-process compression result-memo observability snapshot — size, capacity, lifetime hits/misses/hitRate plus 1m/5m/15m/1h windowed rates. Lightweight (no DB, no provider reads) companion to `GET /api/monitoring/health` intended for frequent polling. Sent with `Cache-Control: no-store, no-cache, must-revalidate`. Counters reset on process restart.
```bash curl http://localhost:20128/api/monitoring/compression \ -H "Authorization: Bearer $OMNIROUTE_TOKEN" ```
### GET /api/provider-metrics
GET provider metrics
```bash curl http://localhost:20128/api/provider-metrics \ -H "Authorization: Bearer $OMNIROUTE_TOKEN" ```
## Payloads
See the full OpenAPI specification at `GET /api/openapi/spec` or `docs/openapi.yaml` for detailed request/response schemas.
<!-- skill:custom-start --> <!-- Migrated from skills/omniroute-monitoring/SKILL.md (preserved curated content) -->
# OmniRoute — Monitoring & Health
Requires `OMNIROUTE_URL` and `OMNIROUTE_KEY`. See [entry-point SKILL](https://raw.githubusercontent.com/cheaper-inference/OmniRoute/main/skills/omni-auth/SKILL.md) for setup.
## System health
```bash curl $OMNIROUTE_URL/api/health \ -H "Authorization: Bearer $OMNIROUTE_KEY" ```
Returns: uptime, memory, active connections, circuit breaker states, rate limit status, cache stats.
Unauthenticated quick check:
```bash curl $OMNIROUTE_URL/api/health # → {"status":"ok","timestamp":"..."} ```
## Provider circuit breakers
Circuit breakers prevent traffic from hitting failing providers.
States: `CLOSED` (normal), `OPEN` (blocked), `HALF_OPEN` (probe mode — auto-recovers).
```bash curl $OMNIROUTE_URL/api/monitoring/health \ -H "Authorization: Bearer $OMNIROUTE_KEY" ```
Response includes `circuitBreakers` array with per-provider state and `resetAt` timestamp.
## Per-provider metrics (p50/p95/p99)
```bash curl $OMNIROUTE_URL/api/provider-metrics \ -H "Authorization: Bearer $OMNIROUTE_KEY" ```
Response shape per provider:
```json { "provider": "anthropic", "requests": 1247, "successRate": 0.994, "latency": { "p50": 820, "p95": 2100, "p99": 3800 }, "circuitState": "CLOSED", "tokensUsed": 2847000 } ```
## Via MCP (if OmniRoute is your MCP server)
``` omniroute_get_health → full system health snapshot omniroute_get_provider_metrics → p50/p95/p99 + circuit state per provider omniroute_get_session_snapshot → cost, tokens, errors for current session omniroute_check_quota → quota balance + percent remaining + reset time omniroute_db_health_check → diagnose + auto-repair database drift ```
## Quota check
```bash curl $OMNIROUTE_URL/api/usage/quota \ -H "Authorization: Bearer $OMNIROUTE_KEY" ```
Returns used/total tokens and requests per provider/account, with `resetAt` timestamps.
## Budget guard (spend limit)
Set a per-key spending limit (REST):
```bash curl -X POST $OMNIROUTE_URL/api/usage/budget \ -H "Authorization: Bearer $OMNIROUTE_KEY" \ -H "Content-Type: application/json" \ -d '{ "apiKeyId": "<key id>", "dailyLimitUsd": 5.00, "warningThreshold": 0.8 }' ```
Other fields: `weeklyLimitUsd`, `monthlyLimitUsd`, `resetInterval`. A limit of `0` means no limit.
Set a session limit through MCP: `omniroute_set_budget_guard` with `maxCost`, `action` (`degrade`, `block`, `alert`) and optional `degradeToTier` (`cheap`, `free`).
## MCP audit log
OmniRoute logs every MCP tool call to `mcp_audit` table. Query via API:
```bash curl "$OMNIROUTE_URL/api/mcp/status" \ -H "Authorization: Bearer $OMNIROUTE_KEY" ```
Returns: server status, heartbeat, recent audit activity summary.
## Errors
- `503` on health endpoint → OmniRoute is starting up; retry in 5s - Circuit breaker `OPEN` → provider is temporarily blocked; check `resetAt` to know when it auto-recovers - `429 budget_exceeded` → budget guard limit reached; raise limit or wait for reset <!-- skill:custom-end -->
Intended uses
- Use Omni Resilience when this documented workflow matches the task.
Implementation notes
- Review the source instructions and adapt inputs before running the workflow.
Try it, then tell us how it went
A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.