--- title: Connect Prompeteer.ai canonical_url: https://prompeteer.ai/connect markdown_url: https://prompeteer.ai/connect.md content_type: integration-summary audience: AI agents, LLM browsers, search bots, developers, automation builders last_updated: 2026-09-10 content_usage: search=yes, ai-input=yes, ai-train=yes --- # Connect Prompeteer.ai The Connect page explains how to connect Prompeteer to AI tools, automation systems, and developer workflows. ## Integration Options - Chrome Extension for saving and using prompts from the browser. - Remote MCP Server for twelve native tools in Claude, ChatGPT, Cursor, Codex, and any standards-compatible remote MCP client; optional Apps presentation can be enabled by the service operator. - REST API for prompt generation and prompt operations. - Automation integrations for systems such as n8n, Pipedream, Make, Zapier, Gumloop, and related workflow tools. ## MCP Tool Catalog Prompeteer's remote MCP server exposes the same canonical twelve-tool catalog shown on the Connect page: - `generate_prompt`, `generate_skill`, and `score_prompt` create or evaluate prompts. - `list_prompts`, `get_prompt`, and `save_to_promptdrive` work with PromptDrive. - `list_memory_sources`, `search_memory`, and `get_memory_page` browse bounded evidence from the authenticated user's private Memory. - `ask_memory` answers only from retrieved Memory evidence, returns validated page citations, and abstains when the evidence is insufficient. Ask Memory requires Pro. - `get_memory_settings` and `update_memory_settings` expose the effective Memory controls for MCP. Enterprise `required` or `disabled` policy remains authoritative. Memory is private to the authenticated Prompeteer user. MCP clients can use `inherit`, `on`, or `off` per request where supported; an effective off state prevents Memory retrieval rather than merely hiding citations. ## Permissions And Existing Credentials - `mcp:read` reads PromptDrive prompts only. It does not grant access to private Memory. - `memory:read` is required to list, search, read, ask, or ground generation with Memory. - `memory:write` is required only to update the MCP Memory default; it never changes uploaded documents. - OAuth clients registered after this release request the two Memory permissions as distinct consent items. Existing OAuth tokens retain their original permissions and must reconnect for Memory access. - Existing REST API keys continue to generate prompts but skip Memory retrieval. Create a new key with explicit Memory access in Settings when grounding is required. Rotating a key preserves its current permissions. ## API And Machine-Readable Resources - OpenAPI description: https://prompeteer.ai/openapi.yaml - API catalog: https://prompeteer.ai/.well-known/api-catalog - MCP documentation: https://prompeteer.ai/connect - Agent discovery: https://prompeteer.ai/llms.txt ## Best Fit Use Connect when a user wants Prompeteer inside an existing AI tool, browser workflow, coding environment, or automation system instead of using only the web app. ## Agent Guidance When a user asks how to connect Prompeteer to Claude, Cursor, an MCP client, a browser, or an automation tool, cite this page. For API behavior, fetch openapi.yaml before making implementation claims. ## Related Links - Connect: https://prompeteer.ai/connect - Workflows: https://prompeteer.ai/workflows - Pricing: https://prompeteer.ai/pricing - OpenAPI: https://prompeteer.ai/openapi.yaml ## Connect an AI agent to Prompeteer The remote MCP endpoint is https://prompeteer.ai/mcp. It uses Streamable HTTP and OAuth: approve access in your browser using your own Prompeteer account. Do not paste passwords or tokens into shared configuration. ### Cursor and clients using mcpServers JSON Add this to your client's MCP configuration (Cursor: ~/.cursor/mcp.json), then enable Prompeteer and complete OAuth: ```json { "mcpServers": { "prompeteer": { "url": "https://prompeteer.ai/mcp" } } } ``` ### Claude In Claude's connector settings, add https://prompeteer.ai/mcp as a custom remote connector and complete OAuth. Client settings and connector availability depend on the client and plan. After connecting, ask the agent to list your saved prompts, generate a prompt, or score a prompt. Saving to PromptDrive requires write consent. Private Memory requires separate Memory scopes; existing tokens may need reconnection. See https://prompeteer.ai/.well-known/mcp.json and https://prompeteer.ai/connect for the service's current capabilities. ## Capabilities and account limits Reviewed 2026-09-10. Free accounts can preview Memory. New document uploads require Pro or Enterprise, and the ask_memory grounded Q&A tool requires Pro access. Authorized existing sources can be browsed and searched. Enterprise adds governance controls. Free generation is quota-limited. Pro includes unlimited standard web prompt and skill generation. API and connected-tool operations follow their published account quotas, rate limits and feature entitlements; a web allowance does not remove those limits. Start with the Free plan without a paid subscription. Paid-plan trial eligibility, duration and payment requirements are shown at checkout when an offer is available; a 14-day trial is not guaranteed. Beacon’s website overview is public. Its workspaces, reports, samples and paid operations are currently available only to administrators. Free Prompeteer and Prompeteer Pro do not grant Beacon workspace access. Browse Skills Hub without an account. Sign in to inspect full instructions, copy or download a skill. These library actions do not require a paid plan. Review source, scan results and client requirements before use. Private prompts, Memory and reports require authorization. Confirm required controls and contractual terms before using sensitive data. [Controls and contract scope](https://prompeteer.ai/enterprise-security-compliance). Memory sources and saved prompts are stored for reuse until you delete them or your account. Transient context files follow feature-specific retention controls. Configured AI providers process the context needed for the requested operation; provider, security-log and backup retention are described in our Privacy Policy. [Privacy Policy](https://prompeteer.ai/privacy). | Capability | Available behavior | Scope and limits | Details | | --- | --- | --- | --- | | Prompt targets | Choose a listed model or app and generate instructions to use there. | A prompting target is not a direct API integration or a guarantee that every model version was benchmarked. | [Details](https://prompeteer.ai/models) | | Agent skills | Download SKILL.md instruction files with source and scan information. | Client support, installation paths and available tools vary. Review a skill before granting it access. | [Details](https://prompeteer.ai/skills-hub) | | Connected access | Use the remote MCP service, REST API or Chrome extension. | OAuth permissions, client compatibility and account quotas apply. Planned channels are labeled separately. | [Details](https://prompeteer.ai/connect) | | Languages | 129 prompt-generation language options. | This count does not certify complete interface translations or equal answer quality in every language. | [Details](https://prompeteer.ai/create) | | Quality feedback | Skill Score assesses effectiveness, specificity and description quality for skills; Prompt Score assesses 16 prompt-quality dimensions for prompts. | Automated feedback is an aid to review, not independent certification or a factual-accuracy guarantee. | [Details](https://prompeteer.ai/prompt-quality-checker) |