CLI tool that summarizes text or articles at a URL using LLM. Supports Anthropic (Claude), OpenAI (GPT), Google (Gemini), and local Ollama models.
uv tool install git+https://github.com/realpacific/summarizerRun the interactive wizard to choose a provider, model, and API key:
summarizer init
#? Choose a provider: Anthropic (Claude)
#? Choose a model: claude-haiku-4-5
#? Enter your Anthropic (Claude) API key: ****************Re-run it anytime to switch providers or models.
# Summarize a URL
summarizer https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete
# Summarize inline text
summarizer "some text to summarize"
# Summarize from stdin
cat file.txt | summarizer
# Override the model for one run
summarizer --model claude-haiku-4-5 https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete
# Choose how the summary is shaped
summarizer https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete --style=brief
# Enter a follow-up chat with the content
summarizer https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete --ask
# For dynamic pages or paywalled content, copy to clipboard and pipe:
pbpaste | summarizer--style controls how the summary is shaped. All styles output plain text with no markdown,
so they stay readable in a terminal and pipe cleanly into other tools.
| Style | Output |
|---|---|
default |
Title, key facts, then the main insight. Used when --style is omitted. |
brief |
One paragraph, three sentences at most — what it is and the single biggest takeaway. |
detailed |
Every substantive argument as its own paragraph, plus any caveats the content raises. |
key-points |
A numbered list of five to ten takeaways. |
summarizer https://example.com/long-article --style=brief
summarizer https://example.com/long-article --style=key-points--ask enters a follow-up chat with the content.
summarizer https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete --ask
# Ask a follow-up question (Ctrl+C or Ctrl+D to quit)
# > what is the key takeaway of this
# The key takeaway is that files written for AI agents like AGENTS.md, GEMINI.md, or CLAUDE.md have become
# better sources of project documentation than the traditional README.md. This is because developers are forced to
# be more detailed, precise, and thorough when writing instructions for machines in order to get the best results
# from AI agents. Other machine-readable files like package management files and CI/CD pipeline configs
# also serve as reliable, up-to-date sources of project understanding. Pair it with realpacific/readthis to turn any summary into on-demand audio.
Install realpacific/readthis:
uv tool install --python 3.12 git+https://github.com/realpacific/readthisPipe any summary directly to it:
# Summarize a URL and read it aloud
summarizer https://prashantbarahi.com.np/blog/your-readme-md-is-obsolete | readthis
# For dynamic pages or paywalled content, copy to clipboard and pipe:
pbpaste | summarizer | readthis