Skip to content
← Tools / Skills
Skill · .md packagen0an/10 trustMetrics-checked
Worknotes logo

Worknotes

One home for everything an agent writes during a session

Visit official site →

About

Worknotes is a small MIT-licensed skill that gives every working document an agent produces — investigations, root-cause write-ups, plans, drafts, review notes — one home: <repo-root>/_agent/docs/. Each document links back to the session that produced it (claude --resume id), folders that grow get an index naming the canonical document, and superseded findings get a banner saying what still holds instead of being deleted. It keeps _agent/ out of git via .git/info/exclude and survives throwaway worktrees by resolving the main checkout. Install with npx skills add https://github.com/n0an/worknotes-skill --skill worknotes. Thin-vetting caveat: 1 star and 2 commits since September 2026 — a genuinely useful convention with almost no track record; the skill itself is tiny and auditable, which limits the risk.

What it can do

  • Single _agent/docs/ home for session documents
  • Session links in every document header
  • Index of past documents with status rows
  • Superseded banners preserving evidence trails
  • Git-excluded via .git/info/exclude
  • Worktree-safe via main-checkout resolution

Setup

Run npx skills add https://github.com/n0an/worknotes-skill --skill worknotes and pick your agents. The skill triggers automatically when the agent is about to write a session work product.

Prerequisites

Node.js for the Skills CLI.

Use cases

agent working notesinvestigation write-upssession documentationworktree-safe notes

Works with

Claude CodeCodexCursorgemini-cli

Trust score — 5/10

/10 trustMetrics-checked

Scored 5/10 (metrics tier, metrics-v1). Evidence: Live GitHub page read this session: 1 star, MIT license, 2 commits; install command copied verbatim from README.; _agent/docs/ convention, session links, superseded banners, git-exclusion, and worktree safety all verified in README.; Very thin vetting: 1 star, 2 commits since September 2026 — metrics tier; skill is tiny and auditable, which bounds the risk..

★ 1 GitHub stars

Sources

Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.