# Portkey MCP connector

The Portkey connector brings your prompts, prompt versions, collections, labels and partials to Claude, ChatGPT, Cursor, Microsoft Copilot and any MCP client, so your team can find, edit and test prompts in plain language once Portkey is connected in Elaichi.

Source: https://elaichi.ai/connectors/portkey/

## Facts

| | |
| --- | --- |
| Application | Portkey |
| Category | Artificial Intelligence |
| AI tools | 239 |
| Authentication | Connects with an API key |
| Bring your own OAuth app | No |
| MCP endpoint | https://api.elaichi.ai/mcp |
| Works with | Claude, ChatGPT, Cursor, Microsoft Copilot, any MCP client, and the Elaichi Agent |
| Tools advertised by name | No. Connected tools are never listed one by one, however few there are. The endpoint advertises `search_tools` and `execute_tool` instead |

## What you can ask once Portkey is connected

- Which Portkey prompts changed in the last week?
- List Portkey collections and the labels on each.
- Summarize completion volume across our Portkey collections.

## Connect Portkey in Elaichi

This happens once for the organization, before any client is involved.

1. Open Connections, choose Add connection, and pick Portkey.
2. Optionally set Share with, then press Connect.
3. Paste a Portkey API key. One person generates a token in Portkey and pastes it once. Everyone else works through Share with, and never sees it.

Credentials are vaulted and nobody, including the AI, reads them back. The connection becomes a toolbox immediately, so you can curate which Portkey tools are exposed, rename them, or freeze arguments before anyone points a client at it.

## Portkey MCP connector for Claude

Endpoint: https://api.elaichi.ai/mcp

1. Open Customize, then Connectors.
2. Press Add.
3. Name it, paste the MCP server URL, then Continue.
4. Sign in and approve.

On Team and Enterprise, an Owner adds it once. Everyone else turns it on for themselves.

## Portkey MCP connector for Claude Code

Endpoint: https://api.elaichi.ai/mcp

1. In a terminal, run `claude mcp add` with the endpoint.
2. Start Claude Code and run `/mcp`.
3. Pick elaichi, then sign in and approve in the browser.

```bash
claude mcp add --transport http elaichi --scope user https://api.elaichi.ai/mcp
```

Use `--scope project` instead to share it with the repository through `.mcp.json`.

## Portkey MCP connector for ChatGPT

Endpoint: https://api.elaichi.ai/mcp

1. Open Plugins, then press the + button.
2. Name it and paste the endpoint into Server URL.
3. Leave Authentication on OAuth, then tick the risk acknowledgement.
4. Press Create, then sign in and approve.

Works on the web. The plugin directory lives at chatgpt.com/plugins.

## Portkey MCP connector for Cursor

Endpoint: https://api.elaichi.ai/mcp

1. Open `~/.cursor/mcp.json`.
2. Add the endpoint under `mcpServers`.
3. Reload Cursor, then sign in and approve.

Set up per machine, so repeat it on each computer you work from.

## Portkey MCP connector for Microsoft Copilot

Endpoint: https://api.elaichi.ai/mcp

1. In Copilot Studio, open your agent, then Tools, Add a tool, New tool and Model Context Protocol.
2. Paste the endpoint into Server URL, choose OAuth 2.0, then Dynamic discovery, and press Create.
3. Create a new connection, sign in and approve, then press Add to agent.
4. Under Channels, open Teams and Microsoft Copilot and make the agent available in Microsoft Copilot.

Microsoft Copilot was called Microsoft 365 Copilot. Keep the tool on end-user credentials, so each person signs in as themselves.

## Connect Portkey to any MCP client

Endpoint: https://api.elaichi.ai/mcp

1. Add the endpoint as a remote MCP server.
2. Sign in and approve.

The Elaichi Agent already has these tools, with nothing to set up.

## What the consent screen decides

Every scope the client asks for starts ticked except "Delete things", which you tick yourself. Calling a Portkey tool needs "Use your connected apps". Over MCP there is no trusted place to confirm a write in the moment, so the consent screen is the standing approval rather than a formality. Think hard before granting Use your connected apps, which can also delete in connected apps and cannot be undone.

## What teams do with Portkey through Elaichi

### Find the prompt behind a feature

AI team. Ask which Portkey prompt powers the onboarding assistant and read its current text, without clicking through the dashboard.

### Tighten a prompt after user feedback

Product. Describe the change you want, review the new prompt version, and set it as the default once you are happy with it.

### Keep prompts organized by collection

Platform. Move prompts into the right Portkey collections and add labels such as production or experiment so the catalog stays tidy as it grows.

### Test a prompt with a real completion

Data science. Run a chat completion through Portkey with a sample question and compare the answer against the previous version before rolling it out.

### Check which prompt version is live

Support. Confirm which version of a Portkey prompt customers are hitting right now when an odd answer gets reported, then decide whether to roll back.

### Reuse shared text with prompt partials

Engineering. Create a Portkey partial for the standard safety instructions once, then reference it from every prompt instead of pasting the same paragraph around.

## Elaichi vs Zapier MCP vs Composio for Portkey

All three can connect Portkey to an AI assistant, and all three have admin controls. They differ in where access lives and how you pay.

| What to check | Elaichi | Zapier MCP | Composio |
| --- | --- | --- | --- |
| Where the AI connects | One address for the whole organization. Endpoint: https://api.elaichi.ai/mcp | A server per member, created at sign-in. | An MCP endpoint per team, or an SDK. |
| Control over Portkey tools | Allow or restrict single Portkey tools, per role or user. | App and action restrictions on the account. | Role permissions, down to the action. |
| Record of calls | One audit entry per Portkey call. | A History tab of tool calls. | A log of every tool call. |
| Single sign-on | SAML or OIDC, plus SCIM, on Gold. | SAML on Enterprise. | SAML and OIDC on Enterprise. |
| Price | Free for one person; $15 per user per month on Gold. | 2 tasks per successful call. | Billed per tool call. |

Sources: Zapier MCP [docs](https://docs.zapier.com/mcp/get-started/quickstart), [security](https://docs.zapier.com/mcp/manage/security), [usage](https://docs.zapier.com/mcp/features/usage); Composio [docs](https://docs.composio.dev/docs/composio-connect), [gateway](https://composio.dev/mcp-gateway), [enterprise](https://composio.dev/enterprise), [pricing](https://composio.dev/pricing). Checked September 2026.

Longer take: [Zapier MCP alternative](/blog/zapier-mcp-alternative/) and [when you don't need an MCP gateway](/blog/when-you-dont-need-an-mcp-gateway/).

## Frequently asked questions

### How do I connect Portkey to Claude?

Connect Portkey in Elaichi by pasting a Portkey API key, which takes about a minute. Then in Claude open Customize, then Connectors, then Add, and paste https://api.elaichi.ai/mcp. There is no OAuth application to register and no client ID or secret to generate, and Claude will ask you to sign in to Elaichi as yourself.

### Does Portkey work with ChatGPT and Cursor as well as Claude?

Yes. Once Portkey is connected in Elaichi, the same endpoint, https://api.elaichi.ai/mcp, works in Claude, ChatGPT, Cursor, Microsoft Copilot, any MCP client and the Elaichi Agent. You connect Portkey once and every client picks it up.

### What can an AI agent actually do with my Portkey data?

With Portkey connected, an agent can list your prompts, read a specific prompt or version, create and edit prompts, set which version is the default, and organize prompts into collections and labels. It can also create partials for shared text and run chat completions through Portkey to test a prompt. Short, concrete asks such as "show me the live version of the refund prompt" work better than long instructions.

### Does connecting Portkey give the AI my whole Portkey workspace?

No. Every call to Portkey runs inside the access of the person who signed in, so the agent can only see the prompts, collections and labels that person can already reach in Portkey. Elaichi can narrow that further, for example to reading only, and it never widens what someone could do on their own.

### Can my team share one Portkey connection?

Yes. One person connects Portkey in Elaichi and shares it with a team, and nobody else ever handles the Portkey API key. Each teammate still signs in to Elaichi as themselves, so the audit log names the actual person behind every prompt change.

### Can I stop an agent from deleting or changing things in Portkey?

Yes. Restrictions in Elaichi work per action, so you can allow reading Portkey prompts and collections while restricting deletes or edits. A restricted action still appears in Elaichi's tool search, flagged as restricted, but Claude, ChatGPT or any other client cannot run it, whatever the prompt says.

### What happens to a Portkey connection when someone leaves?

Offboarding that person in Elaichi ends their access to Portkey through every client at once. If they had shared a Portkey connection with a team, it keeps working for everyone else. When you no longer want Portkey connected at all, disconnecting it once in Elaichi removes it from Claude, ChatGPT, Cursor, Microsoft Copilot and every other client.

### Does the Portkey MCP connector work with Gemini, Codex, Claude Code or other MCP clients?

Yes. Portkey is reached over the same MCP endpoint every client uses, so anything that speaks MCP can call it — Gemini, Codex, Claude Code, Windsurf, Cline, Zed and OpenCode among them — alongside Claude, ChatGPT, Cursor, Microsoft Copilot, and the Elaichi Agent. The tools on offer and the access behind them are identical whichever client asks. Only the setup screen differs.

### Is Elaichi an alternative to Zapier MCP for Portkey?

Yes. Both let Claude, ChatGPT or Cursor use Portkey. Zapier MCP fits a team that already automates in Zapier, since each person signs in and acts as themselves in that account. Elaichi fits when IT wants one address for the whole company, per-tool rules by role, and a record of every Portkey call.

### How is Elaichi different from Composio for Portkey?

Composio gives AI agents tools and sign-in handling across 1,000+ apps, for developers building agents or people using an assistant, billed per tool call. Elaichi gives a company's own people governed access to Portkey: one address, restrictions per role or user, and $15 per user per month. Both have role permissions and a log of every call.

## All 239 Portkey tools

Every tool below is callable through https://api.elaichi.ai/mcp once Portkey is connected, subject to the toolbox it is in and the restrictions on the caller.

- **Create a Portkey chat completion** (Create). Create a chat completion in Portkey by sending a conversation's messages and a model name. Returns: id, choices, created, model, system_fingerprint, object, usage. Required: messages, model.
- **List all Portkey realtimes** (List). Connect to the Portkey Realtime API WebSocket endpoint for streaming realtime sessions. Returns a WebSocket upgrade (101) response on success; the subsequent streaming session follows the provider's realtime protocol. The model query parameter is often required for OpenAI-style realtime sessions.
- **Create a Portkey completion** (Create). Create a text completion in Portkey by sending a prompt to a model via the AI gateway. Returns: id, choices, created, model, system_fingerprint, object, usage. Required: model, prompt.
- **List all Portkey collections** (List). List collections in a Portkey workspace, optionally filtered by name search. Returns each collection with id, slug, name, workspace_id, and parent_collection_id. Required: workspace_id.
- **Create a Portkey collection** (Create). Create a new collection in a Portkey workspace. Returns: id, slug. Required: name.
- **Get single Portkey collection by ID** (Get). Get a single Portkey collection by id. Returns the collection object including id, slug, name, workspace_id, and parent_collection_id. Required: id.
- **Update a Portkey collection by ID** (Update). Update a Portkey collection's name by id. Returns: id. Required: name.
- **Delete a Portkey collection by ID** (Delete). Delete a Portkey collection by id. Returns an empty response on success.
- **List all Portkey labels** (List). List labels in Portkey, optionally filtered by organisation, workspace, or name search. Returns: id, name, description, color_code, organisation_id, workspace_id, is_universal, created_at, last_updated_at, status, updated_at. Max 100 per page.
- **Create a Portkey label** (Create). Create a new label in Portkey. Returns the created label object including id, name, created_at, and updated_at. Required: name.
- **Get single Portkey label by ID** (Get). Get a single Portkey label by id. Returns: id, name, organisation_id, workspace_id, created_at, updated_at. Required: id.
- **Update a Portkey label by ID** (Update). Update an existing Portkey label by id. Returns the updated label object including id, name, created_at, and updated_at. Required: id.
- **Delete a Portkey label by ID** (Delete). Delete a Portkey label by id. Returns an empty 204 response on success. Required: id.
- **List all Portkey prompts** (List). List prompts in Portkey, optionally filtered by collection, workspace, or search term. Returns: id, slug, name, collection_id, model, status, created_at, last_updated_at, object, version_id.
- **Create a Portkey prompt** (Create). Create a new prompt in Portkey. Returns: id, slug, version_id, object. Required: name, collection_id, string, parameters, virtual_key.
- **Get single Portkey prompt by ID** (Get). Get a single Portkey prompt by id or slug. Returns: id, slug, name, collection_id, string, prompt_version, prompt_version_id, prompt_version_status, prompt_version_description, prompt_version_label_id, virtual_key, model, functions, tools, tool_choice, template_metadata, is_raw_template, status, created_at, last_updated_at, version_id, object. Required: id.
- **Update a Portkey prompt by ID** (Update). Update a Portkey prompt's metadata and/or create a new version with updated template content. Set patch to true for partial version updates where omitted fields are inherited from the latest version. Returns: id, slug, prompt_version_id. Required: id.
- **Delete a Portkey prompt by ID** (Delete). Delete a Portkey prompt by id. Returns an empty 204 response on success.
- **List all Portkey prompt versions** (List). List all versions of a prompt in Portkey. Returns: id, slug, version_id, object, name, workspace_id, organisation_id, status, owner_id, updated_by, created_at, last_updated_at, config, format. Required: prompt_id.
- **Update a Portkey prompt version by ID** (Update). Update the label of a specific prompt version in Portkey. Only label_id can be updated; version content (string, parameters, model) is immutable. Returns: id, slug, version_id, object, name. Required: prompt_id, id.
- **Get single Portkey prompt version by ID** (Get). Get a specific version of a prompt in Portkey by version id. Returns: id, slug, name, collection_id, string, prompt_version, prompt_version_id, prompt_version_status, prompt_version_description, prompt_version_label_id, virtual_key, model, functions, tools, tool_choice, template_metadata, is_raw_template, status, created_at, last_updated_at, version_id, object, version_description, total, data.…
- **Portkey prompt make defaults bulk update** (Update). Set a specific version of a Portkey prompt as the default version. Returns an empty 200 response on success. Required: prompt_id, version.
- **List all Portkey prompts partials** (List). List prompt partials in Portkey, optionally filtered by collection. Returns each partial's id, slug, and version_id. Optional: collection_id.
- **Create a Portkey prompts partial** (Create). Create a new prompt partial in Portkey. Returns: id, slug, version_id. Required: name, string. Optional: workspace_id (required for Admin keys), version_description.
- **Get single Portkey prompts partial by ID** (Get). Get a single prompt partial in Portkey by its id or slug. Returns: id, slug, version_id. Required: id.
- **Update a Portkey prompts partial by ID** (Update). Update a prompt partial in Portkey by id. Returns: prompt_partial_version_id. Required: id.
- **Delete a Portkey prompts partial by ID** (Delete). Delete a prompt partial in Portkey by id. Returns an empty 204 response on success. Required: id.
- **List all Portkey partial versions** (List). List all versions of a prompt partial in Portkey. Returns: version_id, version, version_description, string, created_at, updated_at, id, name, workspace_id, slug, organisation_id, status, owner_id, updated_by, last_updated_at, config, format, object. Required: prompt_partial_id.
- **Portkey partial make defaults bulk update** (Update). Set a specific version as the default for a prompt partial in Portkey. Returns an empty object on success confirming the default version was set. Required: prompt_partial_id, version.
- **Create a Portkey prompt completion** (Create). Execute a saved prompt template in Portkey to generate completions with variable substitution. Returns: status, headers, body. Required: prompt_id, variables. Hyperparameters must be passed at the root level alongside variables; max_tokens is deprecated — use max_completion_tokens instead.
- **Create a Portkey prompt render** (Create). Render a Portkey prompt template by substituting its variable values. Returns: success, data. Required: prompt_id, variables. Hyperparameters such as max_completion_tokens and temperature are passed at the root level, not nested under hyperparameters. The max_tokens parameter is deprecated — use max_completion_tokens.
- **List all Portkey guardrails** (List). List guardrails in Portkey, optionally filtered by workspace or organisation. Returns: id, name, slug, organisation_id, workspace_id, status, created_at, last_updated_at, owner_id, updated_by, checks, actions, updated_at. Max 100 per page.
- **Create a Portkey guardrail** (Create). Create a new guardrail in Portkey with specified checks and actions. Returns the created guardrail including id, slug, name, checks, and actions. Required: name, checks, actions.
- **Get single Portkey guardrail by ID** (Get). Get a specific Portkey guardrail by id (UUID or slug with guard_ prefix). Returns: id, slug, name, workspace_id, organisation_id, checks, actions, created_at, updated_at. Required: id.
- **Update a Portkey guardrail by ID** (Update). Update an existing guardrail's name, checks, or actions in Portkey. Returns the updated guardrail object including id, name, slug, checks, actions, organisation_id, and workspace_id. Required: id.
- **Delete a Portkey guardrail by ID** (Delete). Delete an existing guardrail in Portkey by id (UUID or slug). Returns an empty 204 response on success. Required: id.
- **Create a Portkey images generation** (Create). Create an image in Portkey by generating it from a text prompt with a specified model. Returns: b64_json, url, revised_prompt. Required: prompt.
- **Create a Portkey images edit** (Create). Create an image edit in Portkey from an original image and a text prompt describing the desired changes, with an optional mask for targeted edits. Returns: b64_json, url, revised_prompt. Required: image, prompt.
- **Create a Portkey images variation** (Create). Create variations of an existing image in Portkey. Returns the image variation response including created timestamp and a data array of generated image objects. Required: image.
- **Create a Portkey embedding** (Create). Create text embeddings through Portkey's AI Gateway. Returns: index, embedding, object. Required: model, input.
- **Create a Portkey rerank** (Create). Rerank documents by semantic relevance to a query using Portkey's unified interface to reranking models from Cohere, Voyage, Jina, Pinecone, Bedrock, and Azure AI. Returns the rerank response. Required: model, query, documents.
- **Create a Portkey ocr** (Create). Extract text and structured content from documents (PDFs and images) using OCR models in Portkey, supporting providers including Mistral AI and Azure AI Foundry. Returns: pages, model, usage_info. Required: model, document.
- **Create a Portkey audio speech** (Create). Generate speech audio from text via Portkey's AI gateway. Returns a binary audio stream (application/octet-stream) with chunked transfer encoding — write the response body to a file to obtain the audio. Required: model, input, voice.
- **Create a Portkey audio transcription** (Create). Create an audio transcription in Portkey by uploading an audio file with a specified model. Returns: text. Required: file, model.
- **Create a Portkey audio translation** (Create). Create an audio translation in Portkey by submitting an audio file to a speech-to-text model for translation to English. Returns: text. Required: file, model.
- **List all Portkey files** (List). List files in Portkey, optionally filtered by purpose. Returns: id, bytes, created_at, filename, object, purpose, status, status_details. Max 100 per page.
- **Create a Portkey file** (Create). Upload a file to Portkey for use across Assistant, Fine-Tuning, and Batch endpoints. Returns: id, bytes, created_at, filename, object, purpose, status, status_details. Required: file, purpose. Bucket limit 100 GB.
- **Get single Portkey file by ID** (Get). Get information about a specific file in Portkey by id. Returns: id, bytes, created_at, filename, object, purpose, status, status_details. Required: id.
- **Delete a Portkey file by ID** (Delete). Delete a file in Portkey by id. Returns: id, object, deleted. Required: id.
- **List all Portkey contents** (List). Get the contents of a file in Portkey. Returns the raw file content as a string. Required: file_id.
- ...and 189 more tools. Call `tools/list` via the MCP endpoint, or see the full catalog via the API, for the complete set.
