checkrd
how checkrd compares

How Checkrd compares.

Checkrd is a control plane for AI agents. Observability tools record what happened. Gateways route between providers. We do something different. Here’s the honest table, and the cases where another tool is the better choice.

three questions

Which tool do you actually need?

Most teams buy this category for one of three jobs. Pick yours and skip the matrix.

Question 1

Do you need to block a call before it leaves your app?

If yes:

Checkrd

Observability tools watch; gateways route. Only Checkrd enforces, in-process.

Question 2

Do you need to route between LLM providers or cache responses?

If yes:

AI gateway (Cloudflare, Portkey, Kong)

Gateways live at the edge. Many teams run a gateway and Checkrd: routing at the edge, governance in the app.

Question 3

Do you need to iterate on prompts, evals, or datasets?

If yes:

LLM observability (Langfuse, Helicone, LangSmith)

Purpose-built for prompt iteration. Pair with Checkrd: they analyze trace content, Checkrd enforces and audits.

Block requests before they go out

LLM observability
×
AI gateway
×
Checkrd

Trace every call

LLM observability
AI gateway
Checkrd

Kill switch (sub-second cloud, polled air-gap)

LLM observability
×
AI gateway
×
Checkrd

Cryptographically signed telemetry batches

LLM observability
×
AI gateway
×
Checkrd

Rate limit by request body (per model, per user)

LLM observability
partial
AI gateway
rarely
Checkrd

Stores prompts and completions for replay

LLM observability
AI gateway
×
Checkrd
×by design

Sees prompts in transit

LLM observability
AI gateway
Checkrd
×never

Routes between LLM providers

LLM observability
×
AI gateway
Checkrd
×

Caches responses

LLM observability
×
AI gateway
Checkrd
×

Evals, datasets, prompt management

LLM observability
AI gateway
×
Checkrd
×

Runs inside your application

LLM observability
×
AI gateway
×
Checkrd

Open-source core

LLM observability
varies
AI gateway
varies
Checkrd

Air-gap deployment

LLM observability
×
AI gateway
×
Checkrd
when an alternative is right

Two cases where you should pick something else.

when llm observability is the right call

Langfuse · Helicone · LangSmith · Phoenix · Braintrust

If your work is iterating on prompts, comparing models, or building evaluation suites, an observability tool is purpose-built for that. Many teams run both. Checkrd enforces policy and keeps the audit log; the observability tool analyzes trace content where retention is acceptable.

when an ai gateway is the right call

Cloudflare AI Gateway · Portkey · Kong AI plugins

If you want provider routing, response caching, or caller-level rate limiting at the edge, an AI gateway is purpose-built for that. Checkrd is a different shape: a governance layer inside your application, not a routing fabric. The two work well together. The gateway handles routing and caching, Checkrd handles policy, audit, and the kill switch.

not for everyone

When Checkrd isn’t the right fit.

A shorter list than most vendor sites publish. Worth reading before you install.

Single-agent prototypes

If you have one agent and you haven't shipped to production yet, a hard-coded allowlist in your application is probably enough. Come back when you have two agents, two teams, or a compliance question to answer.

Pure prompt-iteration workflows

If your day is writing, diffing, and evaluating prompts, an observability platform fits the job better. Checkrd doesn't replace eval tooling.

Zero compliance surface

If you have no CISO, no auditor, no regulated customers, and no plans to add any of the three, governance is probably not what you need today.

Still the right fit?

The SDK is on PyPI and npm. Email us if your deployment needs VPC, air-gap, or a vendor-review packet.

$ pip install checkrd