Skip to main content

b2c-edge-traffic-triage

Investigate bot activity, scrapers, crawlers or abusive traffic hitting a B2C Commerce storefront through the eCDN (CDN edge). Review WAF, firewall and rate-limiting protections, correlate available traffic evidence, and propose narrowly scoped rule changes for approval.

Informations de source

Dépôt
SalesforceCommerceCloud/b2c-developer-tooling
Dernière activité de la source
7 octobre 2026 à 16:34
Langue détectée de SKILL.md
anglais
Étoiles
56
Forks
22

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
b2c-edge-traffic-triage
description
Investigate bot activity, scrapers, crawlers or abusive traffic hitting a B2C Commerce storefront through the eCDN (CDN edge). Review WAF, firewall and rate-limiting protections, correlate available traffic evidence, and propose narrowly scoped rule changes for approval.
# Edge traffic and bot triage Outcome: identify which automated traffic matters, whether current edge protections cover it, and a proposed, approved next step. Inspection never authorizes blocking, challenging, or changing zone settings. ## Scope and impact Resolve the instance with `config_inspect` (CLI: `b2c setup inspect`). List the tenant's zones first and focus on production storefront zones; ask when the target zone is ambiguous. Use a fixed UTC window (propose the last 7 days if none is given) and record the symptom: origin load or errors, scraping, login or token abuse, basket/inventory abuse, or a routine review. Record impact and owner; follow the team's incident process for live customer impact. ## Access and tools | Evidence | Preferred MCP path | CLI fallback | | ------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------- | | Zones | Code mode `getZonesInfo` (`cdn/zones/v1`) | `b2c ecdn zones list` | | Edge protections | Code mode reads of security settings, custom firewall rules, rate-limiting rules, managed WAF rulesets | `b2c ecdn security get`, `b2c ecdn firewall list`, `b2c ecdn rate-limit list`, `b2c ecdn waf ...` | | Edge traffic attribution | Discover read operations for edge traffic or analytics with `scapi_search` (`api: "cdn/zones/v1"`, prefer `schemas: "live"`); use only what the contract offers | Logpush jobs (`listLogpushJob`; `b2c ecdn logpush jobs list`) show where raw edge logs already go | | Edge volume and errors | `metrics_get` with `category: "ecdn"` where the Metrics API is enabled | `b2c metrics ecdn` | | Origin impact | `logs_list_files` / `logs_get_recent` | `b2c logs get` | Read `skill://mcp/b2c-mcp-scapi/SKILL.md` before code mode; the zone APIs need `sfcc.cdn-zones` (writes `sfcc.cdn-zones.rw`). If no traffic operation is available to this tenant or client, say that edge attribution is unavailable through the toolkit and continue with protections, Logpush and origin evidence. Never infer traffic volume from rule configuration alone. ## Checks and decisions 1. Keep executions small: one or two zones per `scapi_execute`, fixed `from`/`to`, small result limits. Executions stop at 30 s; split work rather than retry a timed-out batch whole. 2. Classify sources by user agent, ASN, IP, host, path, status and cache result: verified crawlers and link previews, monitoring or integrations, unverified automation (headless browsers, HTTP libraries, empty user agents), and scanning or exploit probes. A user agent is self-reported; corroborate it. 3. Before proposing action on a high-volume source, ask who runs it. Synthetic monitors, performance tests, feeds and Salesforce services often look like bots. 4. Prioritize traffic that reaches origin or sensitive paths (login, token, basket, checkout, search). Mostly cached automation has lower origin cost. 5. Compare sources with protections: rules in a non-enforcing mode (monitor, simulate, log), disabled managed rules, thresholds above observed rates, and rules scoped to other hosts or paths. Report gaps, not only matches. ## Mitigation and recovery Propose the narrowest change (host, path, ASN or verified-bot condition) and a non-enforcing or challenge step before a block. Changes need explicit approval and are subject to Safety Mode. After a change, re-check the same evidence over a fresh equivalent window and keep a rollback (previous rule state). Rule syntax and CLI options: native `b2c-ecdn`, MCP `skill://b2c-cli/b2c-ecdn/SKILL.md` and its `references/SECURITY.md`. ## Escalation and handoff Summarize zones, window, top sources with counts and share, the evidence source and its limits, matching rules and gaps, proposed changes, and next owner. Route suspected attacks with customer impact to the security owner and Salesforce Support. Use native `b2c-production-triage` or the MCP [case template](skill://b2c-ops/b2c-production-triage/references/escalation.md).
Voir sur GitHub