cowpath
Menu
Install

Every process your team repeats gets better each time.

Most work is the same work, done again. You correct your agent, the chat ends, and the fix is gone. cowpath keeps one playbook per process, shared with your team. Every run starts from it, and every correction makes it better.

curl -fsSL cowpath.sh/install | sh
Set up a team

Free while in beta. Works with Claude Code, Codex, claude.ai and ChatGPT.

The good version lives in last week's chat.

Every new chat starts from zero, so you make the same corrections again.

  • Claude

    You · Mon 9:12

    Again: no superlatives. Use their numbers, not ours.

  • Slack · #sales

    Priya · 10:41

    does anyone have the prompt dana used for the acme deck? it was in some chat in august

  • Notion

    Sales prompts (v3 FINAL)

    Last edited 214 days ago by Tom (deactivated)

Two corrections this week. None the next.

cowpath keeps the run, writes the playbook, and loads it the next time the process comes up.

Demo: release notes, corrected twice, then right the next week

  1. user: Write the release notes for v2.4 from the merged PRs.
  2. agent: v2.4 is our most powerful release yet. We've added blazing-fast CSV export, squashed a pesky sign-in bug, and much more.
  3. user: No superlatives. Group under Added and Fixed, one line per change.
  4. agent: Added - CSV export for reports Fixed - Sign-in loop after password reset - Timezone shift in scheduled exports
  5. user: Link the PR numbers. Customers click those.
  6. agent: Added - CSV export for reports (#804) Fixed - Sign-in loop after password reset (#812) - Timezone shift in scheduled exports (#815)
  7. cowpath: run kept · eng/release-notes · 2 corrections
  8. cowpath: playbook v1 written
  9. cowpath: Group under Added and Fixed. One line per change.
  10. cowpath: No superlatives. Say what changed.
  11. cowpath: Link every PR number.
  12. cowpath: next week · new chat
  13. cowpath: playbook loaded · eng/release-notes
  14. user: Release notes for v2.5.
  15. agent: Added - Saved filters on the reports page (#841) Fixed - Duplicate rows in CSV export (#838)
  16. cowpath: 0 corrections

Every process gets its own playbook.

Your team shares one playbook per process. A new hire's agent starts from it on day one.

sales/decks/discovery

41 runs

Five reps, five decks.Now: Every deck opens with the customer's numbers.

Discovery deck

Open#

  • Open with their numbers from the call notes, not ours.

Middle#

  • Product tour: three slides, after the problem.

Close#

  • End on one next step with a date.

Your transcripts stay on your machine.

Only the run leaves: the start, your corrections and the final version. Local mode sends nothing.

One command. Then cowpath init.

In claude.ai and ChatGPT, add https://cowpath.sh/mcp as a connector.

  • curl -fsSL cowpath.sh/install | sh
  • brew install cowpath-sh/tap/cowpath
  • npx cowpath

Questions

What counts as a process?
Work you repeat that ends in something you would make again: a deck, a reply, release notes. One-off questions are skipped.
Isn't this just memory?
No. Memory keeps facts about one person, out of sight. cowpath keeps one playbook per process that your team can read, edit and revert.
Do I have to change how I work?
No. Hooks in Claude Code and Codex keep each run; in claude.ai and ChatGPT your agent opens and closes runs itself.
What does it cost?
Free while in beta. Local mode stays free.