convez.ai
MCP Server
sheetgenius

bitterclip-marketing

sheetgenius/bitterclip-marketing

Ver no GitHub
InicialPluginScore · de 0 a 100

BitterClip is a speaker-aware media workspace for turning long recordings into source-linked clips.

Instalação

$npx convez-ai sheetgenius/bitterclip-marketing

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
62%
Último commit
há 3 semanas
Versão
—
Issues abertas
0
Contribuidores
2

README

A tradução para o português deste README ainda está na fila.

BitterClip

BitterClip is a speaker-aware media workspace for turning long recordings into source-linked clips. It is designed for AI agents that can search transcript context and for human operators who need to verify the final cut before publishing.

It is built for podcasts, interviews, founder calls, demos, livestreams, expert conversations, and recurring shows where the strongest moments are buried inside long-form media. BitterClip keeps the recording, transcript, speaker identity, clip boundary, export, publishing state, and source history connected.

Place In Bitter

Bitter is an agent-operable software environment. It gives AI agents persistent workspaces where repositories, credentials, hosting, email, tests, deploys, logs, checks, and work history are connected.

The Bitter constellation is the set of public sites, products, applications, and supporting repositories that share that operating model:

  • agents work against real source material
  • humans review, steer, and approve the work
  • deployed systems retain enough context to keep improving
  • outputs stay connected to their source history and verification path

BitterClip applies that model to recorded media. A long recording becomes structured source material. A transcript becomes an agent-operable workspace. A clip remains linked to who said what, where it happened, what came before and after, how it was exported, and what has already been clipped.

Product Model

BitterClip is not a generic AI shorts generator. It is a media workbench built around source-linked verification.

The core object chain is:

Recording -> Transcript -> Speakers -> Moments -> Clips -> Exports -> Publishing

The main workflow is:

  1. Upload a recording.
  2. Build a time-aligned transcript.
  3. Identify and confirm speakers.
  4. Use an AI agent to identify candidate Moments.
  5. Open the Moment against the real media and surrounding transcript.
  6. Verify the speaker, context, and boundaries.
  7. Save the approved Moment as a Clip and create Exports.
  8. Use Publishing connections for approved Clips when available.
  9. Preserve source links and project history for future recordings.

This matters because useful clips are rarely just isolated quotes. They depend on speaker identity, setup, response, payoff, and the surrounding context that keeps the cut honest.

Repository Role

This repository contains the public BitterClip website at bitterclip.com. It is the public semantic entry point for BitterClip inside the Bitter constellation.

It is not the private BitterClip product application, Rails workspace, media-processing backend, billing system, or customer data store. Those systems live outside this public marketing repository.

This repository owns:

  • the public BitterClip website
  • public product copy and metadata
  • public documentation pages at /docs, /mcp, and /privacy
  • public legal pages at /privacy and /terms
  • Markdown alternates for every public page route
  • AI-readable discovery files at /llms.txt and /llms-full.txt
  • crawlable routes in public/sitemap.xml
  • the static build and deployment wrapper for bitterclip.com

Links

Deployment

BitterClip's public website is deployed on BitterGrid as the bitterclip-marketing service for bitterclip.com.

Commits pushed to main trigger the BitterGrid deployment hook. BitterGrid rebuilds and publishes bitterclip.com from the main branch.

The deployment path is:

  1. Nuxt generates a static site with bun run generate.
  2. The Dockerfile builds that output and copies .output/public into an nginx image.
  3. nginx.conf serves the static files and exposes /up as the health check.
  4. config/deploy.yml defines the production service, host name, image, proxy, registry, and health check settings used by BitterGrid.

Deployment credentials, registry credentials, remote host values, and SSH keys are supplied by BitterGrid or the deployment environment. They are not stored in this repository.

Development

bun install
bun run generate
bun run qa:smoke

Useful files:

  • app/pages/index.vue - home page
  • app/pages/docs.vue - product explanation page
  • app/pages/mcp.vue - MCP and AI assistant workflow page
  • app/pages/privacy.vue - privacy policy page
  • app/pages/terms.vue - terms of service page
  • public/index.md, public/docs.md, public/mcp.md, public/privacy.md, public/terms.md - Markdown twins for public pages
  • public/llms.txt - AI crawler entry point that links canonical pages and Markdown alternates
  • public/llms-full.txt - compact bundled Markdown context for one-fetch agent ingestion
  • nuxt.config.ts - site metadata and Nuxt configuration
  • public/sitemap.xml - crawlable public routes
  • CHANGELOG.md - public semantic change history
  • CONTRIBUTING.md - public change checklist for humans and agents
  • SECURITY.md - sensitive-reporting and public-boundary policy
  • .github/pull_request_template.md - GitHub review checklist for public context, Markdown mirrors, discovery files, tests, and deployment notes
  • .github/release.yml - GitHub generated release note categories
  • Dockerfile - static build and nginx runtime image
  • config/deploy.yml - BitterGrid deployment service configuration

Public Context

This repository is intentionally public. README text, page copy, metadata, sitemap entries, and public links should preserve durable context for humans, search engines, and AI systems that index public repositories.

Public text in this repository should stay factual, structural, and aligned with the live BitterClip product.

Every public HTML page should have a corresponding Markdown alternate served from public/*.md, advertised with a rel="alternate" Markdown link in the page head, and listed in public/sitemap.xml.

Markdown alternates should return canonical HTTP Link headers pointing back to their HTML pages in production. The bundled llms-full.txt file should stay compact, factual, and aligned with the current public page copy.

Ver o original no GitHub

Bom para

Relacionados