helipod-mcp 0.5.6 → 0.5.7

package/dist/http.js

@@ -38076,7 +38076,7 @@ var generalTools = [
 ];
 
 // src/server.ts
-var VERSION = "0.5.6";
+var VERSION = "0.5.7";
 var INSTRUCTIONS = `Helipod is a cloud platform (Indonesia): containers, managed Postgres databases, VPS, domains, email and an AI gateway, billed from a prepaid wallet in Rupiah.
 - Start with list_projects to find projects and environment ids. Account tools (get_wallet, invoices, statements, notifications, get_project_activity, support tickets) cover the account; AI tools (list_ai_models, get_ai_overview, get_ai_usage, AI keys, top-ups) cover the AI gateway.
 - Anything that costs money (create/deploy/scale services, resize volumes, upgrade a database plan, create/renew/upgrade VPS, VPS backups, register/renew/transfer domains, email plans and inbox add-ons, AI top-ups and auto top-up): call it with dryRun=true first (or the price lookup when it has no dry run), tell the user the cost in IDR and the wallet balance after, and wait for an explicit yes before the real call. Never treat silence or an earlier general OK as a yes.

package/dist/index.js

@@ -37221,7 +37221,7 @@ var generalTools = [
 ];
 
 // src/server.ts
-var VERSION = "0.5.6";
+var VERSION = "0.5.7";
 var INSTRUCTIONS = `Helipod is a cloud platform (Indonesia): containers, managed Postgres databases, VPS, domains, email and an AI gateway, billed from a prepaid wallet in Rupiah.
 - Start with list_projects to find projects and environment ids. Account tools (get_wallet, invoices, statements, notifications, get_project_activity, support tickets) cover the account; AI tools (list_ai_models, get_ai_overview, get_ai_usage, AI keys, top-ups) cover the AI gateway.
 - Anything that costs money (create/deploy/scale services, resize volumes, upgrade a database plan, create/renew/upgrade VPS, VPS backups, register/renew/transfer domains, email plans and inbox add-ons, AI top-ups and auto top-up): call it with dryRun=true first (or the price lookup when it has no dry run), tell the user the cost in IDR and the wallet balance after, and wait for an explicit yes before the real call. Never treat silence or an earlier general OK as a yes.

package/dist/stdio.js

@@ -37328,7 +37328,7 @@ var generalTools = [
 ];
 
 // src/server.ts
-var VERSION = "0.5.6";
+var VERSION = "0.5.7";
 var INSTRUCTIONS = `Helipod is a cloud platform (Indonesia): containers, managed Postgres databases, VPS, domains, email and an AI gateway, billed from a prepaid wallet in Rupiah.
 - Start with list_projects to find projects and environment ids. Account tools (get_wallet, invoices, statements, notifications, get_project_activity, support tickets) cover the account; AI tools (list_ai_models, get_ai_overview, get_ai_usage, AI keys, top-ups) cover the AI gateway.
 - Anything that costs money (create/deploy/scale services, resize volumes, upgrade a database plan, create/renew/upgrade VPS, VPS backups, register/renew/transfer domains, email plans and inbox add-ons, AI top-ups and auto top-up): call it with dryRun=true first (or the price lookup when it has no dry run), tell the user the cost in IDR and the wallet balance after, and wait for an explicit yes before the real call. Never treat silence or an earlier general OK as a yes.

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "helipod-mcp",
   "mcpName": "io.helipod/helipod",
-  "version": "0.5.6",
+  "version": "0.5.7",
   "description": "MCP server for Helipod: containers, databases, VPS, domains and email from Claude Code, Codex, Cursor and other AI tools.",
   "type": "module",
   "bin": {

package/skills/helipod-ai-gateway/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: helipod-ai-gateway
-description: Add AI features to an application using the Helipod AI Gateway (ai.helipod.io) - OpenAI and Anthropic compatible endpoints, one key for many models, billed per token in Rupiah. Use when the user wants to call LLMs from their code through Helipod or switch an OpenAI/Anthropic integration to Helipod.
+description: Add AI features to an application using the Helipod AI Gateway (ai.helipod.io) - OpenAI and Anthropic compatible endpoints plus image generation, one key for many models, billed per token in Rupiah. Use when the user wants to call LLMs from their code through Helipod or switch an OpenAI/Anthropic integration to Helipod.
 ---
 
 # Helipod AI Gateway
@@ -38,6 +38,18 @@ ai = OpenAI(base_url="https://ai.helipod.io/v1", api_key=os.environ["HELIPOD_AI_
 - Prices per 1M tokens (input / output / cached) are on https://helipod.io/ai-gateway. Some models cost more at peak hours (Mon to Fri 08:00 to 11:00 and 13:00 to 17:00 WIB).
 - Small/fast for classification and chat; larger models for reasoning and code.
 
+## Images
+
+`POST https://ai.helipod.io/v1/images/generations` (OpenAI images format, same key): `{ model, prompt, size?, n?, images?: [https urls], response_format?: "url"|"b64_json", async?: true }`.
+
+```ts
+const img = await ai.images.generate({ model: "seedream-image-v5.0-flash", prompt: "A wooden boat at sunrise, watercolor" });
+```
+
+- Image models are NOT in `/v1/models`; they are listed with sizes and prices at https://docs.helipod.io/docs/ai-gateway/images (cheapest: `seedream-image-v5.0-flash`, about Rp 320 per image).
+- **Images are kept for 7 days, then deleted with their links** (`expires_at` in every answer). When an app needs images longer, download them right away and store them in the app's own storage; never save only the returned URL. Tell the user this when you build it.
+- The request waits up to ~80 s; otherwise it answers `202` with an `id`: poll `GET /v1/images/generations/{id}` until `status` is `succeeded` or `failed`. Failed images are not charged.
+
 ## Practical rules
 
 - Stream (`stream: true`) for long answers and reasoning models. One request may take at most 10 minutes; set the client timeout to at least 10 minutes. Failed requests are not charged.