Your Status Page Is an Agent Readiness Signal
Before an AI agent delegates work to your API, it asks one question: are you up? D8 Reliability carries a 0.13 weight — the second-highest dimension in the Agent Readiness Score. A /health endpoint takes 15 minutes to add and impacts 13% of your score.
Why Agents Check Your Status Before Delegating Work
When a human user asks an AI agent to “process this payment through Stripe” or “send this email via Resend,” the agent has a choice. It can blindly call the API and hope it works. Or it can check the service's status first, confirm it is operational, and proceed with confidence.
Well-designed agents always check. The cost of checking is one HTTP request — typically under 50ms. The cost of not checking is a failed delegation that the agent has to debug, retry, or escalate to the user. A /health endpoint that returns 200 OK is the green light. A 503 Service Unavailable tells the agent to try another service or wait. No endpoint at all means the agent is flying blind.
This is exactly what D8 Reliability measures. Not just whether your service is up right now, but whether you communicate your operational status in a way that agents can consume programmatically. The distinction matters: a service with 99.99% uptime but no status page scores lower on D8 than a service with 99.9% uptime and a comprehensive status infrastructure.
What AgentHermes Checks for D8 Reliability
The D8 Reliability dimension scans for five signals, from the most machine-readable (a /health endpoint) to the most human-readable (SLA documentation). Each adds points.
/health endpoint
Weight: HighA JSON endpoint returning 200 when operational and 503 when degraded. The simplest machine-readable status check. Agents call this before every delegation.
Example: GET /health → { "status": "ok", "uptime": "99.98%", "latency_ms": 42 }
status.domain.com
Weight: HighA dedicated status page (Statuspage, Instatus, Betterstack, or self-hosted) showing component-level health, incident history, and scheduled maintenance.
Example: status.stripe.com shows API, Dashboard, and Webhook health independently.
/status page
Weight: MediumAn alternative to the subdomain approach. A /status path on the main domain serving the same information. AgentHermes checks both locations.
Example: yoursite.com/status → component statuses, recent incidents, uptime percentage.
Incident history
Weight: MediumPast incidents with timestamps, root cause, impact, and resolution time. Agents use historical reliability data to assess risk before delegating critical tasks.
Example: Last 90 days: 2 incidents, avg resolution 14 minutes, 99.97% uptime.
SLA documentation
Weight: LowPublished Service Level Agreement with uptime guarantees, response time commitments, and compensation terms. The contractual layer that agents use for critical delegations.
Example: 99.99% uptime SLA, <200ms p95 response time, credits for breaches.
The Three-Tier Approach: From 15 Minutes to Full SLA
You do not need to implement everything at once. Each tier builds on the previous one and adds incremental score value.
Tier 1: /health (Machine-Readable)
15 minutesA single JSON endpoint that returns 200/503 with basic metrics. This is what agents actually call before delegating. It is the minimum viable status signal.
{ "status": "ok", "version": "2.1.0", "uptime": "99.98%", "checks": { "database": "ok", "cache": "ok", "queue": "ok" } }
Score impact: Adds 3-5 points to D8. Moves businesses from "unknown reliability" to "programmatically verifiable."
Tier 2: Status Page (Human + Machine)
30 minutesA dedicated page showing component-level health, incident history, and scheduled maintenance. Use Statuspage, Instatus, Betterstack, or a simple static page.
Components: API (Operational), Dashboard (Operational), Webhooks (Degraded). Incidents: 2 in last 90 days.
Score impact: Adds 5-8 points to D8. Provides the context agents need for risk assessment: not just "up now" but "how reliable historically."
Tier 3: SLA Doc (Contractual)
1-2 hoursPublished SLA with uptime guarantees, response time commitments, support tiers, and compensation policy. This is the trust layer for critical agent delegations.
Uptime: 99.99%. Response: <200ms p95. Support: 15min response (critical). Credits: 10x for breach.
Score impact: Adds 2-3 points to D8. Unlocks trust for high-value delegations where agents need contractual guarantees.
Start with Tier 1.A single /health endpoint returning JSON takes 15 minutes. It is a route in your web framework that checks your database connection, returns 200 if healthy and 503 if not. That alone moves you from “unknown reliability” to “programmatically verifiable” — and adds 3-5 points to D8.
Who Does It Right: Every Silver Scorer Has a Status Page
This is not a coincidence. Every business that scores Silver (60+) on the Agent Readiness Score has all three tiers: a /health endpoint, a public status page, and published SLA documentation. It is table stakes for the top tier.
The pattern is universal. Resend (75 Gold), Vercel (70 Silver), Supabase (69 Silver), Stripe (68 Silver), and GitHub (67 Silver) all have all three tiers. They all have a /health endpoint that returns JSON, a dedicated status subdomain with component-level monitoring, and a published SLA. The businesses that score below 40 almost never have any of the three.
This creates a clear roadmap. If you want to enter Silver territory, you need status infrastructure. It is not the only requirement — D2 API Quality matters more — but without D8 Reliability, you are leaving 13% of the possible score on the table.
Implementation: Add a /health Endpoint in 15 Minutes
Here is the minimum viable /health endpoint for common frameworks.
Response format
Return JSON with status ("ok" or "degraded"), uptime percentage, version string, and component checks (database, cache, external dependencies). HTTP 200 for healthy, 503 for degraded.
Component checks
Each dependency gets its own status. If your database is down but your cache is up, return 503 with database: "down", cache: "ok". This lets agents make informed decisions about which operations are safe.
No authentication
The /health endpoint must be publicly accessible. An agent checking your status before delegation should not need credentials. This is the one endpoint where no-auth is correct.
Low latency
The health check should return in under 100ms. A slow health check defeats its purpose — agents need a fast yes/no before proceeding. Check connectivity, not query performance.
Cache headers
Set Cache-Control: no-cache, no-store. Status must be real-time. A cached "ok" from 5 minutes ago is worse than no status at all if the service went down since.
After the /health endpoint, add a status page. Free options: Instatus (free tier with 1 status page), Betterstack (free monitoring + status page), or a simple static HTML page you update manually. Link it from your homepage footer and from your agent-card.json. The status page URL is one of the fields AgentHermes checks during D8 scoring.
What Happens Without a Status Page
Agents skip the reliability check
Without a /health endpoint, agents delegate blindly. If your service is down, the agent discovers the failure only after the request fails — wasting time, budget, and user trust.
No incident context
When an agent encounters an error, it cannot distinguish between "this service is having an outage" and "my request was malformed." Without a status page, every failure looks like the agent's fault.
D8 scores near zero
Without any status infrastructure, D8 Reliability contributes almost nothing to your score. That is 13% of the total score effectively zeroed out — the equivalent of missing an entire dimension.
Agents prefer competitors
Given two services with similar functionality, an agent will prefer the one it can verify is operational. A competitor with a /health endpoint gets chosen over you every time.
Frequently Asked Questions
How much does a status page actually impact my score?
D8 Reliability carries a 0.13 weight — the second-highest dimension after D2 API Quality (0.15). A complete status page setup (/health endpoint + status page + SLA) can add 8-13 points to D8. Since D8 is 13% of the total score, that translates to roughly 1-2 points on your overall Agent Readiness Score. It sounds small, but in a scoring system where the difference between Bronze and Silver is 20 points, every dimension matters.
Do I need a paid status page service?
No. A /health endpoint is free — it is a single JSON route in your application. A basic status page can be a static HTML page you update manually or a free-tier service like Instatus or Betterstack. The paid services (Statuspage by Atlassian, Betterstack Pro) add automated monitoring, subscriber notifications, and incident management workflows. Agents only need the machine-readable data, which the free tier provides.
What should my /health endpoint return?
At minimum: HTTP 200 with a JSON body containing a status field ("ok" or "degraded"). Better: include component checks (database, cache, queue, external dependencies), uptime percentage, current version, and response latency. Best: add a machine-readable schema (JSON Schema or OpenAPI response definition) so agents can programmatically parse every field. Return 503 when any critical component is down.
Why do agents care about status pages?
Agents delegate work to APIs. Before delegating a critical task (processing a payment, booking an appointment, sending a notification), a well-designed agent checks if the service is operational. A /health endpoint answers that question in milliseconds. Without one, the agent either skips the reliability check (risky) or tries a test request and interprets the result (slow and fragile). Status pages are the difference between a confident delegation and a guess.
Does your business have a status page?
Run a free Agent Readiness Scan to see your D8 Reliability score and find out if agents can verify your operational status.