Your AI Writes the Code. gitglimpse Writes the Standup.

The new development loop

Modern development increasingly involves AI tools. Claude Code writes your functions, Cursor suggests your implementations, Copilot completes your lines. These tools produce better commit messages, more structured changes, and more frequent commits than most developers write manually. But at the end of the day, you still need to communicate what you accomplished. gitglimpse bridges that gap.

Three ways to use gitglimpse with AI tools

Template mode reads your commit messages and file changes instantly, no LLM needed. This works great when your AI coding tool already writes descriptive commit messages — and most of them do. For richer summaries, add a local LLM with --local-llm and gitglimpse will read your actual code diffs, understanding what changed at a deeper level than commit messages alone. Or pipe --json output into Claude Code or Cursor and let the editor's built-in AI format your standup.

# Template mode — instant, offline
glimpse standup

# With local LLM — reads code diffs for richer output
glimpse standup --local-llm --context both

# JSON for Claude Code / Cursor
glimpse standup --json

The /standup slash command

Run glimpse init in your project and commit the generated files. Every developer who pulls the repo gets /standup, /report, /pr, and /week as slash commands in Claude Code. Type /standup and Claude reads the structured JSON output, formats a polished summary, and presents it right in your terminal. No copy-pasting, no context switching.

PR summaries are where LLMs shine

The glimpse pr command generates pull request descriptions by analyzing what your branch changed compared to main. In template mode, you get a structured list of changes and stats. With an LLM, it reads every line of your diff and writes a coherent paragraph describing what the entire branch accomplishes — not just listing commits, but synthesizing intent. This is especially powerful when your commits were generated by an AI tool that may have written technically accurate but contextually sparse messages.

# PR summary — defaults to --context both
glimpse pr

# With LLM for richer narrative
glimpse pr --local-llm

Why this matters

• •AI tools produce more commits with better messages — gitglimpse gets better input automatically.
• •The /standup slash command turns a daily chore into a one-word command.
• •PR summaries with LLM mode read actual diffs, producing descriptions you'd actually want on a pull request.
• •Everything works offline and privately — your code never leaves your machine unless you choose a cloud LLM.