Skip to content
Back to skills

Walnut Inbox Triage

ASecurity

The procedure for ONE Inbox Triage run: what to read, which notes to update, and what to ask. Use only when launched as the Inbox Triage agent with a batch of new mail and Slack items. Not for one message, not for triaging tasks (the `walnut-triage` skill), not for general mail or Slack questions.

  • 32 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 1, 2026
ai-agents

Security analysis

A100/100

Scanned October 1, 2026

npx -y skills add EvanZhang008/open-walnut --skill walnut-inbox-triage --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Walnut Inbox Triage?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Walnut Inbox Triage
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/evanzhang008-walnut-inbox-triage/badge)](https://www.skillsdirectory.com/skills/evanzhang008-walnut-inbox-triage)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: walnut-inbox-triage
description: >-
  The procedure for ONE Inbox Triage run: what to read, which notes to update,
  and what to ask. Use only when launched as the Inbox Triage agent with a batch
  of new mail and Slack items. Not for one message, not for triaging tasks (the
  `walnut-triage` skill), not for general mail or Slack questions.
---

# One triage run

One batch, worked once, then stop.

## Order

1. Read `notes/Walnut/Triage/State.md` (its body rides the batch when present).
2. Read only what the batch omits: `mail_read`, `slack_*`, `task_search`,
   `task_list`.
3. Match each item to a project and tasks that already exist.
4. Update the project's tracking note: `project_tracking_get`, then
   `project_tracking_ensure` if it has none.
5. Ask the tasks affected: `task_send` with `expect_reply`.
6. Write the letters (below).
7. Then rewrite `State.md`, append a line to
   `notes/Walnut/Triage/Runs/<YYYY-MM>.md`, and `memory_write` what lasts (a
   sender's meaning, a channel → project route, a preference).

## Letters (`human_inbox_send`)

- **One** summary: `review`, with `task_refs` for every task you touched.
- **Three** decisions at most: `action_required`, each with buttons —
  `Make a task` / `Reply for me` / `Unsubscribe` / `Ignore`. No buttons is
  refused, and so is a 4th: fold the rest into the summary, and do not retry.
- A tap comes back into THIS run, even hours later. Act, then `human_inbox_reply`.
  `Reply for me` means a DRAFT plus `mail_request_send`, never a send.
- Withdraw a decision an earlier run left behind once its item is handled.

## The notes

- `notes/Walnut/Triage/State.md` — rewrite every run: `## Awaiting` (rq-/lt- ids),
  `## Watching`, `## Recently handled`, `## Notes for next run`.
- `notes/Walnut/Triage/Runs/<YYYY-MM>.md` — append one line per run.
- `notes/Projects/<P>/Tracking.md` — per item: `## Status`, `## Workstreams`,
  `## Open questions`, `## Log`. A `## Workstreams` edit needs the `content_hash`
  from a `note_read` in THIS run, or it is refused. A `## Log` append targets the
  heading anchor.

## Never

- Never send mail, post to Slack, mark anything read, or unsubscribe yourself:
  ask with `mail_request_send`, `slack_request_post`, `mail_unsubscribe_request`.
  Your run instructions say what `assist` adds; those three never change.
- Never invent a project, and never create a task for something the user has not
  shown they care about; if you are unsure, say so in the summary.
- Never re-handle an item under `## Recently handled`.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.