Neomutt Commander

PassAudited by ClawScan on May 1, 2026.

Overview

This instruction-only skill is coherent for managing email with neomutt, but it needs IMAP credentials and can change mailbox state.

Before installing, make sure you are comfortable letting the agent access the configured mailbox. Prefer a revocable app password, keep the neomutt config file protected, and require confirmation before sending email or making large mailbox changes.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Installing and using this skill may give the agent access to read and manage the configured email account through neomutt.

Why it was flagged

The skill requires email-provider credentials that can grant mailbox access. This is expected for neomutt/IMAP, but it is high-impact account access and is not reflected in the registry's primary credential declaration.

Skill content
set imap_user = "user@gmail.com"
set imap_pass = "*****"
... For Gmail, `imap_pass` should be an **App Password**
Recommendation

Use a revocable app password where available, protect the neomutt config file, and only configure accounts you are comfortable letting the agent access.

What this means

Reading or organizing messages can mark them read, move them, archive them, or create/send drafts depending on user direction.

Why it was flagged

The neomutt workflows can persistently change mailbox state and can send email, although the artifact documents these actions and requires approval before sending.

Skill content
`Automatic` | Message is marked read when opened ... `s` ... Move ... `y` | **Send** (only if sending is approved)
Recommendation

Give explicit instructions for which messages may be modified, and require confirmation before any send, bulk move, deletion, or archive operation.

What this means

A malicious or misleading email could try to influence the agent if the agent treats message text as commands rather than content to summarize or act on only with approval.

Why it was flagged

Email bodies and rendered HTML are external content that may enter the agent's context. This is purpose-aligned, but email content should not be treated as trusted instructions.

Skill content
Press `Enter` on a message. HTML parts render automatically via w3m.
Recommendation

Treat email content as untrusted data, and ask for user confirmation before following requests, links, attachments, or instructions contained in messages.