generated: '2026-08-26'
method: searched
source: https://api.vynaris.com/agents.md
name: Vynaris Agent Skills
description: >-
  Vynaris publishes its own agent-facing operations manual at
  https://api.vynaris.com/agents.md and links it from llms.txt as "the complete
  agent-facing manual - read this". It is saved here verbatim rather than
  generated, because the provider authored it: it carries the onboarding flow,
  the behavior contract, error semantics, harness configuration for six named
  harnesses, and precise pricing an agent can compute with. No skill in this
  index was authored by API Evangelist.
skills:
- file: vynaris-agents.md
  name: vynaris-operations-runbook
  title: Vynaris agent onboarding and operations runbook
  method: searched
  source: https://api.vynaris.com/agents.md
  fetched: '2026-08-26'
  http_status: 200
  content_type: text/plain; charset=utf-8
  bytes: 12023
  provider_last_updated: '2026-08-22'
  api: Vynaris Gateway API
  operations:
  - POST /v1/accounts
  - GET /v1/ping
  - POST /v1/chat/completions
  - GET /v1/models
  - GET /v1/usage
  - GET /v1/ledger
  covers:
  - programmatic self-onboarding (POST /v1/accounts returns a provisional key, $0 balance)
  - connectivity and key-state bisection via GET /v1/ping
  - first request against model "auto"
  - the response cost receipt (vynaris block + x-vynaris-* headers)
  - error semantics including the terminal, do-not-retry 402
  - harness configuration for OpenAI SDK, Anthropic Python SDK, Claude Code, Cursor, Hermes, OpenClaw and Pi
  - a stated "cases where Vynaris may not fit" section
notes:
- >-
  There is no OpenAPI for this provider, so operations above are the endpoint
  strings the provider documents in prose, not operationIds resolved from a
  machine-readable contract.
- >-
  No API-Evangelist-generated skills were added. The provider-published runbook
  is a stronger artifact than anything this pipeline could derive without a spec.