generated: '2026-09-04'
method: generated
source: openapi/webscale-networks-webscale-apis-openapi.json (Webscale APIs 2026.273)
name: Webscale Networks agent skills
api: Webscale APIs
provider_published_skills: none-found
note: >-
  Searched first for provider-published skills or an AGENTS.md on webscale.com, docs.webscale.com and
  both GitHub orgs (github.com/webscale-networks, github.com/section-io) — none exists. These skills
  are GENERATED by API Evangelist from the provider's own published OpenAPI, and every step is grounded
  in a real method+path pair from that contract. Because the contract declares no operationId on any
  of its 151 operations, steps are identified by method + path rather than by an invented id.
skill_count: 2
skills:
- file: skills/webscale-networks-deploy-a-web-application.md
  name: webscale-deploy-a-web-application
  flow: onboard a storefront onto the Webscale platform
  operations: 7
  destructive_steps: false
  covers_recovery: true
- file: skills/webscale-networks-manage-edge-access-control.md
  name: webscale-manage-edge-access-control
  flow: allow/block traffic via address sets and audit live web controls
  operations: 12
  destructive_steps: true
  covers_recovery: true
safety_notes_baked_into_every_skill:
- No idempotency mechanism exists — read-back-and-reconcile is prescribed instead of blind retry.
- No dry-run/preview parameter exists on any operation.
- No reversal operation or documented reversal window exists for any of the 74 mutating operations.
- No 4xx/5xx responses are declared, so error branching is explicitly discouraged.
- No API rate limit or RateLimit-* header is published; the platform's customer-facing rate-limiting feature is called out as a different thing.
- Web controls are read-only in the published contract; skills say so rather than implying writes.