Open Source · MIT Licensed

Your terminal is now a
marketing workbench.

11 Claude Code skills that read your product docs and generate content for technical audiences. No SaaS. No dashboard. Just a git repo that makes Claude Code write like your engineering team talks.

$ git clone github.com/j1ngg/tech-marketing-framework
$ cd tech-marketing-framework
$ claude
Ready in 30 seconds

Config-as-code marketing

This is not an app. It is a git repo. Clone it, fill in six markdown files about your product, and Claude Code becomes a senior product marketer who knows your product, your competitors, and your audience.

Every skill reads from docs/inputs/ before generating anything. The output is grounded in your actual product, not generic AI copy full of "seamless" and "cutting-edge."

Product context in, accurate marketing out. No hallucinated features. No fabricated quotes. No placeholder links.

docs/inputs/
├── product_brief.md
├── target_personas.md
├── messaging_positioning.md
├── competitor_intel.md
├── testimonials.md
└── brand_guidelines.md

docs/inputs-local/  (gitignored, private override)

11 skills and growing

Each one is a slash command. Each one reads your product docs first.

/social-posts

LinkedIn and Twitter posts from your product docs

/email

Event follow-up sequences with programmatic personalization

/blog

SEO and AEO optimized posts with tri-publish variants

/ads

Google, Meta, and LinkedIn ad copy via creative matrix

/sales-deck

B2B narrative decks using April Dunford methodology

/image

Marketing images via MCP with brand guideline enforcement

/messaging-positioning

Socratic workshop that builds your positioning doc

/editorial-calendar

Monthly rolling content plan using MKT1 methodology

/launch-roundup

MRs to Slack + blog + LinkedIn + carousel in one run

/skill-builder

Design and create new skills or agents interactively

/autoresearch

Autonomous eval loops that optimize your skill prompts

3 agents and growing

Fire and forget. No interaction required.

asset-reviewer

Asset Reviewer

Ruthless QA agent. Reviews any marketing asset against your content guidelines, product brief, and positioning. Flags banned words, structural anti-patterns, and missing evidence.

how-they-market

Competitive Research

Runs parallel web searches across HN, Reddit, LinkedIn, and Twitter to surface exactly how a company, person, or newsletter does marketing. Outputs a strategic flywheel, channels, tactics, and what to steal.

ads-auditor

Ads Auditor

Analyzes campaign metrics against benchmarks. Produces a health score (0 to 100), critical issue flags, optimization recommendations, and platform-specific checklists.

How it works

1

Clone and fill in your docs

Add your product brief, personas, competitive intel, messaging, and testimonials to docs/inputs/.

$ git clone github.com/j1ngg/tech-marketing-framework
$ cd tech-marketing-framework
2

Run Claude Code

Claude loads the marketing persona, content guidelines, and all skills automatically.

$ claude
3

Use any skill

Every skill reads your product docs before generating. Output references your real product, not hallucinated features.

> /blog
> /social-posts
> /ads

Built-in AI slop killer

The asset-reviewer agent enforces your content guidelines on every piece of output. Banned words, structural anti-patterns, and unsubstantiated claims get flagged before anything ships.

ChatGPT / Generic AI

"Unlock the power of seamless integration with our cutting-edge platform. Leverage robust AI capabilities to transform your workflow and delve into unprecedented productivity gains."

Tech Marketing Framework

"Ship a LinkedIn post about your new vector search feature in 30 seconds. /social-posts reads your product_brief.md, knows your competitors from competitor_intel.md, and writes like your engineering team talks. Not like a SaaS marketing team."

Same prompt. Different context. The difference is your product docs.

Skills that improve themselves

/autoresearch runs your skill against binary evals, scores outputs, mutates the prompt one change at a time, and keeps improvements. Based on Andrej Karpathy's autoresearch methodology.

> Running autoresearch on /email skill...

> Baseline: 23/30 passed (76.7%)

> Mutation 1: Added specific event detail requirements
> Result: 30/30 passed (100.0%)
> Improvement: +23.3%

> New best score achieved. Saving optimized skill.

Real results from the /email skill optimization run.

MIT Licensed

Open source.

Obsessively optimized daily

by a startup marketing VP.

Ready in 30 seconds.

$ git clone github.com/j1ngg/tech-marketing-framework
$ cd tech-marketing-framework
$ claude
Get started on GitHub