Wisprs

co.wisprsPublished by wisprs.coOfficial publisherv0.30.0Updated Nov 1, 2026

Transcribe audio, video and YouTube links, then search, summarize and export your transcripts.

VerifiedStreamable HTTPWeb executableProductivity & WorkflowKnowledge & MemoryMedia & Design

Overview

AI-generated overview

Wisprs transcribes audio, video and YouTube links, then lets an assistant search, summarize, translate and export the resulting transcripts.

What you can do
  • Start transcription from a public HTTPS media URL, or open an authenticated uploader handoff for local or private files
  • Check asynchronous job status and retrieve completed transcripts in cursor-based pages
  • Search the connected library for ranked moments and fetch generated artifacts
  • Summarize transcripts, generate chapters, repurpose them into show notes, threads, blog drafts or quotes, and translate them
  • Export transcripts as TXT, SRT, VTT, JSON, Markdown or DOCX as a private expiring resource
When to use it
Useful when you want an assistant to turn recordings, videos or YouTube links into searchable transcripts and derived content such as summaries, chapters, show notes or translated versions, without moving media files through the model.
Requirements
A remote Streamable HTTP endpoint at wisprs.co, with OAuth browser authorization required; no static bearer tokens are accepted. A Wisprs account and connected tenant are implied. No environment variables or headers are declared.
Before you install
Transcription starts billable asynchronous work, so confirm before starting it and never retry an ambiguous request with a new idempotency key, which can create duplicate charges. OAuth tokens should be completed in the browser and never pasted into prompts or config files. Exports and transcripts are stored as private expiring resources in your Wisprs tenant.

Installation

In SourceWeft

  1. Open Wisprs in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/co.wisprs%2Fwisprs/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "wisprs": {
      "type": "http",
      "url": "https://wisprs.co/api/mcp"
    }
  }
}

README

Wisprs AI Connectors

Official Wisprs distribution package for ChatGPT, Codex, Claude Code, and remote MCP clients. The authenticated Streamable HTTP endpoints are:

  • Claude connector and plugin: https://wisprs.co/api/mcp/claude (STT and other non-TTS workflows; 39 tools).
  • ChatGPT, Codex, and other direct MCP clients: https://wisprs.co/api/mcp (universal surface; 43 tools, including TTS).
  • Browser setup guide: https://wisprs.co/mcp/setup.

The Claude release supports transcript, library, meeting, creator, support, and developer workflows:

  • transcribe_url starts an asynchronous transcription from a public HTTPS media URL.
  • get_private_media_upload_handoff opens the authenticated Wisprs uploader for local or private media without sending file bytes or upload tokens through the model.
  • get_job_status checks the durable operation without creating duplicate work.
  • get_transcript retrieves a completed transcript in bounded, cursor-based pages.
  • get_transcript_artifact retrieves generated content in bounded pages.
  • search_library finds ranked moments in the connected tenant's library.
  • export_transcript creates TXT, SRT, VTT, JSON, Markdown, or DOCX as a private expiring MCP resource.
  • summarize_transcript and generate_chapters create reusable structured artifacts.
  • repurpose_transcript creates show notes, threads, blog drafts, or quote selections.
  • translate_transcript creates a reusable language-specific artifact.

The server owns authentication, authorization, tenant isolation, idempotency, rate limits, cost controls, audit records, and data retention. This repository contains no Wisprs service source code, credentials, or customer data.

Install for Claude Code

Test a local checkout:

bash
claude --plugin-dir ./wisprs-ai-connectors

Or add the public repository as a marketplace after it is published:

text
/plugin marketplace add Wisprs/wisprs-ai-connectors/plugin install wisprs-ai-connectors@wisprs-plugins

In Claude Code, run its /mcp command to complete OAuth in the browser. Do not paste tokens into the repository, prompts, or configuration files.

Install for ChatGPT and Codex

The .codex-plugin/plugin.json manifest uses .mcp-openai.json and the universal endpoint. During private testing, register https://wisprs.co/api/mcp in ChatGPT developer mode, then associate the resulting connection with this package. Public discovery requires separate OpenAI review; the presence of this package does not imply marketplace approval.

Direct MCP configuration

Claude hosts use the checked-in .mcp.json. Other hosts can use .mcp-openai.json for the universal endpoint. Each endpoint returns OAuth protected-resource metadata and drives browser authorization. Static bearer tokens are neither required nor accepted in this package.

Correct asynchronous usage

  1. Tell the user that transcription starts billable asynchronous work and obtain confirmation when the host has not already done so.
  2. Generate one opaque idempotency key for the logical request.
  3. Call transcribe_url once and retain its operationId.
  4. Poll get_job_status no faster than retryAfterSeconds.
  5. After completed, use the returned opaque transcriptionId with get_transcript.
  6. Follow nextCursor exactly and state when the response is partial.

For exports and AI transformations, use one stable idempotency key, poll the returned operation, then read the exact export resource URI or call get_transcript_artifact with the completed result metadata. Never place a large export directly into conversation context when the MCP resource is available.

Never retry an ambiguous transcribe_url call with a new idempotency key. That can create duplicate billable work.

Development

Requirements: Node.js 22 or newer and, for host validation, current Claude Code and Codex tooling.

bash
npm testnpm run validatenpm run package

npm run validate:live-links additionally verifies public policy and support links over HTTPS. Host-specific test cases and release gates are in docs/HOST_COMPATIBILITY.md and docs/REVIEWER_GUIDE.md.

Security and privacy

Read SECURITY.md before reporting a vulnerability and PRIVACY.md for the package's data boundaries. The service policies are published at wisprs.co/privacy and wisprs.co/terms.

Release status

Version 0.28.3 is a pre-submission package. No OpenAI or Anthropic approval is claimed. See CHANGELOG.md and release.json for compatibility and provenance.

License

MIT. See LICENSE.

Source: README.md at commit b055de5

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

1
  1. v0.30.0LatestNov 1, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts