Skip to main content
Category: Launch · Version: 0.0.1 · Author: opendirectory · Compatibility: All targets (claude, codex, opencode, gemini, anti-gravity, openclaw, hermes)
Tell the skill what your product shipped. It writes a polished, living docs/changelog.md entry and hands you a ready-to-use content package: tweet thread, LinkedIn post, email snippet, and one-liner. Run it after every deploy. Over quarters, docs/changelog.md becomes a complete, shareable product history.

Install

What It Does

  • Accepts: free text items, git commits (auto-read), GitHub PRs — any combination
  • Filters noise: merge commits, version bumps, CI/CD, typos excluded automatically
  • Transforms: technical commit language -> user-facing benefit language
  • Categorizes: New / Improved / Fixed / Under the hood
  • Generates a content package: one-liner, tweet thread, LinkedIn post, email snippet
  • Appends a new dated entry to docs/changelog.md (living log, newest entry first)
  • Saves the content package to docs/product-updates/[date]-content.md

Requirements

No other API keys required. Free text + git auto-read work with no configuration.

Setup

How to Use

Include competitor names for richer channel discovery. Include ICP role + pain for accurate signal-tracing.

What Gets Generated

docs/changelog.md (living log)

docs/product-updates/[date]-content.md (content package)

Input Sources

The skill pulls from all three sources automatically: All three can be combined. Overlapping items are deduplicated.

Changelog Entry Format

Each entry follows a consistent structure:
Version labels default to Week of [Month Day, Year]. Detected semver format is preserved and incremented automatically.

Content Rules

  • Tweets: Under 280 chars each, no hashtags, no em dashes, active voice
  • LinkedIn: No markdown formatting, no hashtags, founder voice (“We shipped” not “We are excited to announce”)
  • Email: 50-100 word body, subject line formula, casual and direct
  • One-liner: Max 20 words, covers top 1-2 items

Standalone Script

Run data collection without Claude. Useful for raw item discovery before analysis.

Project Structure

Cost Per Run

  • Git, free text: free, no auth
  • GitHub PRs: free with optional token
  • AI analysis: uses the model running the skill
  • Total: free

License

MIT