TopNotchEuropeMeshAI

Every project deserves a launch πŸš€ We built an agent that generates strategy, posts, and pitch decks, tracks it in SOCIAL.md β€” all from one prompt.

Comment

Inspiration

Every project deserves a launch πŸš€

Across Europe, engineers build strong AI prototypes that never reach users. After hackathons, momentum dies at go-to-market. We wanted a protocol, not a checklist - something versionable, automatable, and agent-native.

We drew inspiration from AGENTS.md (machine-readable execution rules) and CHANGELOG.md (deterministic progress tracking). Marketing should be as reproducible as CI.

What it does

SOCIAL.md (Social Agent Protocol, SAP) is a repository-native launch protocol.

It allows an AI agent to:

  • Analyze a GitHub repo
  • Infer ICP, positioning, and value proposition
  • Generate go-to-market strategy
  • Produce launch posts and campaign content
  • Create an 8-slide investor pitch deck
  • Publish to LinkedIn and Instagram
  • Track engagement and iterate

All steps are committed into SOCIAL.md, making distribution reproducible and rerunnable from a single prompt.

How we built it

  • Claude Code plugin as execution layer
  • Custom skills system (6 domain skills: ICP, positioning, content, pitch, analytics, distribution)
  • Gemini workflow for structured market analysis
  • GitHub as source of truth
  • SOCIAL.md as protocol file (state + tasks + metrics)
  • CLI-compatible design for Claude, OpenAI Codex, and Cursor

Architecture pattern:

Repo β†’ Analysis β†’ Strategy β†’ Assets β†’ Distribution β†’ Metrics
            ↓
         SOCIAL.md (versioned state)

All outputs are deterministic, version-controlled, and diffable.

Challenges we ran into

  • Translating messy repos into structured positioning
  • Automating multi-platform publishing safely
  • Designing a protocol flexible enough for any project
  • Keeping agent autonomy while preserving reproducibility

The main constraint: marketing is ambiguous; engineering systems are not. We needed a bridge.

Accomplishments that we're proud of

  • Defined SOCIAL.md as a reusable open protocol
  • Built agent-driven GTM from raw GitHub repo
  • Generated automated investor-ready pitch deck
  • Implemented reproducible campaign reruns
  • Shipped cross-model compatibility (Claude, Codex, Cursor)

What we learned

  • Distribution can be protocolized
  • Agents need structured state files
  • Version control is the missing layer in AI marketing
  • Builders adopt tools that behave like code, not dashboards

Marketing becomes manageable when it’s treated as infrastructure.

What's next for TopNotchEuropeMeshAI

  • Public spec of SOCIAL.md (open standard)
  • LinkedIn MCP / official API integration
  • Multi-agent orchestration (research β†’ strategy β†’ execution β†’ analytics loop)
  • Benchmarking launch performance across projects
  • SaaS version for hackathons and student incubators

Goal: make project launches deterministic, repeatable, and agent-native.

Built With

  • 11labs
  • claude
  • gemini
  • linkedin-api
  • plugins
  • remotion
  • skills
  • slidev
Share this project:

Updates