---
skill_id: gf-ops-rate-limit-playbook
evaluation: free
quality: agent-decision-grade
functional: true
price_usd_full: 6.49
price_label: $6.49 / unlock
kind: service
category: Agent Ops
challenge: https://lvlltd.com/api/pay?skill=gf-ops-rate-limit-playbook
outline: https://lvlltd.com/skills/gf-ops-rate-limit-playbook/outline.json
runtime_after_unlock: src/index.js createSkill
standards: x402, ERC-7857
completion_pass: functional-v1
---

# Free sample — Rate-Limit Playbook

**Skill ID:** `gf-ops-rate-limit-playbook`  
**Unlock:** $6.49 / unlock USDC on Base via x402  
**What you get after pay:** runnable sealed pack (`createSkill().run`) + workbook + templates  

## Summary

Rate-limit and backpressure playbooks for agent tool calls, public APIs, and webhook bursts with client and edge tactics.

When agents and clients stampede, limits fail chaotically. This playbook defines budgets, 429 handling, queueing, and edge rules so retries do not amplify outages. Free outline; sealed x402 unlock. No revenue guarantees; not legal advice.

## When to use

- Agent tool storms or user bursts trigger 429s and retry storms that make the outage worse.
- You need shared language for per-key, per-IP, and per-tenant budgets across edge and app layers.
- Webhook or unlock traffic spikes and you lack a documented shed vs queue decision.

## Decision-grade outline (free)

1. Define budgets: per-IP, per-API-key, per-tenant, and per-agent-tool limits with burst and sustained rates.
2. Specify client behavior: Retry-After respect, jittered exponential backoff, and circuit breakers for agent loops.
3. Configure edge and app layers: Cloudflare-style rate rules, app middleware, and queue admission for webhooks.
4. Write shed vs queue criteria: what drops first (expensive tools, anonymous traffic) vs what must buffer.
5. Add observability: limit-hit dashboards, top offenders, and a weekly budget review with owners.

## Capabilities

- Multi-dimension rate budget templates for IP, key, tenant, and tool scopes
- Client retry and circuit-breaker guidance that stops agent retry amplification
- Edge-plus-app coordinated limit design with shed-vs-queue criteria
- Limit-hit observability and offender review checklist

## Artifacts after unlock

- `rate-limit-playbook.md`
- `budget-matrix.yaml`
- `retry-circuit-policy.md`

## Detailed description

Rate limits without a playbook become random 429 walls—and agents are especially good at turning walls into retry storms. The Rate-Limit Playbook gives Agent Ops a sealed, practical system for budgets and backpressure. It starts by defining multi-dimension limits: per IP for abuse, per API key for partners, per tenant for fair multi-tenant use, and per agent tool for expensive model or external calls. Burst vs sustained rates are explicit so short spikes do not look like attacks and sustained abuse is not waved through. Client behavior is first-class: honor Retry-After, use jittered exponential backoff, and open circuit breakers when an agent loop is hammering a dead dependency. Edge and application layers are coordinated so Cloudflare-style rules and app middleware do not fight each other, and webhook admission can queue instead of blindly dropping. Shed-vs-queue criteria name what is expendable under pressure (anonymous scraping, optional tools) versus what must buffer (paid unlock paths, signed webhooks). Observability closes the loop with limit-hit metrics, top offenders, and a weekly budget review. Free outline on the listing; full pack unlocks via x402 Base USDC. Operational guidance only—not legal advice, not a DDoS mitigation contract, and no revenue or availability guarantees. Use it so limits protect the system instead of surprising it.

## Functional runtime (paid)

After verified payment the sealed pack includes:

| Path | Role |
|------|------|
| `src/index.js` | `createSkill`, `run`, `plan`, `checklist` |
| `src/runbook.js` | Outline-driven steps with done_when |
| `src/tools.js` | Tool schemas for agent registries |
| `WORKBOOK.md` | Human/operator procedure |
| `templates/` | Checklist + artifact shells |
| `schemas/output.schema.json` | Result contract |

### Agent snippet (after unlock)

```js
import { createSkill } from "./src/index.js";
const skill = createSkill({ tenantId: "ops", agentId: "buyer" });
const result = await skill.run({
  goal: "Execute Rate-Limit Playbook for my catalog",
  context: { skill_id: "gf-ops-rate-limit-playbook" },
  dry_run: false,
});
// result.ok, result.steps, result.artifacts
```

## Pay path

1. Free: this sample + outline.json  
2. `GET https://lvlltd.com/api/pay?skill=gf-ops-rate-limit-playbook` → HTTP 402  
3. Transfer `maxAmountRequired` USDC to `payTo` on Base  
4. `POST /api/pay` with `X-PAYMENT: {"txHash":"0x…","skill":"gf-ops-rate-limit-playbook"}`  
5. Write `sealed_pack.files` to disk or use install-unlocked-pack.mjs  
6. Re-redeem anytime with the same verified txHash  

## Honesty

- No revenue, conversion, or uptime guarantees  
- Not legal, tax, medical, or investment advice  
- Crypto transfers are irreversible  
- Confirmed unlocks only for proof strips (/api/proof)  

*Public `/sealed/gf-ops-rate-limit-playbook/` is teaser-only. Full `src/` is paywalled under locked sealed delivery.*