Backstop Study 2.0 — an open validation study benchmarking Redlynr's cost-escalation guardrail across 1,568 trials and three models. 100% catch rate. 0% false positive rate. All data and methodology public. Executive summary → Full report →
Official Python SDK available — automatically enforces Redlynr decisions before tool execution. pip install redlynr GitHub → PyPI →
One call before every agent action

Know whether to stop before your agent finds out the hard way.

Redlynr is the guardrail your orchestration layer calls before every step — one API, one answer: proceed, slow down, or stop. Cost, chain depth, retries, and request volume, all checked in a single request.

Try 30 calls free See pricing
POST /run
tenant_id: "acme_corp"
agent_id: "agent_42"
chain_id: "chain_abc123"
depth: 0
step_cost: 0.02
proceed
within budget, depth, and volume limits
slow_down · wait 6s
request volume approaching ceiling
stop · retries_exhausted
agent auto-paused, needs manual clear
n8n Integration  ·  Use Redlynr in n8n. Start with 30 free calls or use the paid x402 workflow with your own Base USDC wallet. View n8n Workflows →
Four dimensions, one check

Everything that can make an agent run away, in one call.

Each dimension has a sensible default the moment you start calling Redlynr — no setup required to try it.

01

Cost budget

Tracks spend against a rolling period at both tenant and per-chain level. Stops before the charge happens, not after. Optional soft warning before the hard ceiling.

default: $50 / 24h tenant · optional per-chain
02

Chain depth

Caps nesting level reported by your orchestration layer. Runaway delegation — an agent calling an agent calling an agent — hits a hard ceiling.

default: 10 levels
03

Steps & retries

Limits steps per chain and retries per agent. Exhaust your retries, and Redlynr can auto-pause the agent until you manually clear it.

default: 25 steps · 3 retries
04

Volume pressure

Watches request rate in a sliding window at both tenant and per-agent level. Warns by default as you approach the ceiling — or hard-blocks, if you'd rather it did.

default: warn only, 80% threshold
Set once, call constantly

Configure thresholds once. Call /run on every action.

No dashboard, no login screen — just two calls your code already knows how to make.

ONBOARDING · ONCE

POST /register

Claim your tenant namespace and receive your owner_token. Required before calling /run. Takes one API call — no account, no email, no dashboard.

SETUP · ONCE

POST /policy

Set your thresholds. Skip this and Redlynr uses sensible defaults automatically — nothing to configure to start testing.

EVERY ACTION

POST /run

Your orchestration layer calls this before each step. Redlynr checks all dimensions and answers: proceed, slow_down, or stop.

WHEN NEEDED

POST /pause · /reset · /audit

Clear an auto-paused agent, reset a chain counter, or pull the last 100 decisions. Targeted calls, not a dashboard.

Validated by Backstop
1,568 trials across 28 scenarios, 10 domains, four conditions, and three models.
Backstop is HDGForge's validation study for Redlynr. The methodology, results, amendments, limitations, executive summary, and full technical report are publicly available.
View the Executive Summary → Explore the Research →
Pricing

Priced for something you call constantly.

$0.001 / call
30 free calls — no account, no card, just solve a quick proof-of-work challenge first (details)
Pay-per-call after that — settled in USDC on Base via x402
No subscription — an idle integration costs nothing
Try 30 calls free →