overlay: 1.0.0 info: title: API Evangelist enhancements for the Edgee AI Gateway API version: 1.0.0 extends: openapi/edgee-openapi-original.json x-generated: '2026-08-17' x-method: generated x-source: 'API Evangelist enrichment pipeline; every value below is derived from artifacts in this repo or from a probe recorded in them. The harvested spec is never mutated.' actions: - target: $.info update: x-apievangelist-profile: https://apis.io/provider/edgee/ x-apievangelist-harvested: '2026-08-17' x-apievangelist-source: https://www.edgee.ai/docs/api-reference/openapi.json x-apievangelist-conventions: conventions/edgee-conventions.yml x-apievangelist-errors: errors/edgee-problem-types.yml x-apievangelist-error-codes: errors/edgee-error-codes.yml x-apievangelist-lifecycle: lifecycle/edgee-lifecycle.yml x-apievangelist-data-model: data-model/edgee-data-model.yml x-apievangelist-authentication: authentication/edgee-authentication.yml x-apievangelist-rate-limits: rate-limits/edgee-rate-limits.yml x-apievangelist-plans: plans/edgee-plans-pricing.yml x-apievangelist-mcp: mcp/edgee-mcp.yml x-apievangelist-note: 'This spec covers only the AI Gateway inference plane on https://edgee.io. The Edgee Console API (https://api.edgee.app) is documented in prose and has no published machine-readable contract.' - target: $.info update: x-idempotency-supported: false x-idempotency-note: No idempotency key or header is documented anywhere in the Edgee docs or in this spec; retries re-execute and re-bill. - target: $.info update: x-error-format: custom-json x-error-envelope: 'error.{message,type,code,param}' x-rfc9457: false - target: $.servers update: - url: https://edgee.io description: Edgee AI Gateway x-apievangelist-verified: '2026-08-17' - url: https://api.edgee.ai description: Gateway alias (CNAME global.edgee.io) — GET /v1/models returned HTTP 200 with 230 models on 2026-08-17 x-apievangelist-method: probed - target: $.paths['/v1/models'].get update: x-apievangelist-anonymous: true x-apievangelist-probe: url: https://api.edgee.ai/v1/models http_status: 200 fetched: '2026-08-17' models_returned: 230 rate_limit_headers: none observed - target: $.paths['/v1/chat/completions'].post update: x-apievangelist-agent-skill: skills/edgee-route-a-chat-completion.md x-apievangelist-action-class: acting x-apievangelist-consequence: write x-apievangelist-note: Billable. No idempotency contract, so a retry is a second charge. - target: $.paths['/v1/messages'].post update: x-apievangelist-agent-skill: skills/edgee-migrate-anthropic-traffic.md x-apievangelist-action-class: acting x-apievangelist-consequence: write - target: $.paths['/v1/responses'].post update: x-apievangelist-action-class: acting x-apievangelist-consequence: write - target: $.paths['/v1/compress'].post update: x-apievangelist-agent-skill: skills/edgee-estimate-and-compress-a-payload.md x-apievangelist-action-class: acting x-apievangelist-consequence: read x-apievangelist-note: Compresses a payload without calling any LLM provider, so it does not incur provider token cost. - target: $.paths['/v1/count_tokens'].post update: x-apievangelist-agent-skill: skills/edgee-estimate-and-compress-a-payload.md x-apievangelist-action-class: connected x-apievangelist-consequence: read - target: $.paths['/v1/messages/count_tokens'].post update: x-apievangelist-action-class: connected x-apievangelist-consequence: read - target: $.tags update: - name: Chat description: Chat completion endpoints (OpenAI format) x-apievangelist-capability: inference - name: Messages description: Messages endpoints (Anthropic format) x-apievangelist-capability: inference - name: Responses description: Responses endpoints (OpenAI Responses API format) x-apievangelist-capability: inference - name: Models description: Model management endpoints x-apievangelist-capability: discovery - name: Tokens description: Token estimation endpoints x-apievangelist-capability: accounting - name: Compress description: Standalone token compression endpoint x-apievangelist-capability: optimization