Thunderbird

ReviewAudited by ClawScan on May 1, 2026.

Overview

This appears to be a legitimate Thunderbird mail-search skill, but it can read private local email, account details, and optionally save attachments.

Install this only if you want the agent to inspect local Thunderbird mail. Use the narrowest profile/account/folder/date filters, avoid full-body searches unless needed, treat emails and attachments as untrusted, and review the bundled script because no upstream source or homepage is provided.

Findings (5)

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

Private messages, subjects, recipients, and body previews may be exposed to the agent and any transcript or downstream workflow using the response.

Why it was flagged

The skill intentionally brings local email metadata and message content into the agent response. This is purpose-aligned, but email contents are private and may become part of the chat context.

Skill content
By default, summarize: source mailbox path, date, from, to, subject, body preview
Recommendation

Use explicit profile, account, folder, date, and limit filters; use full-body output only when necessary.

What this means

A malicious or misleading email could try to tell the agent to change goals or reveal information if the agent treats message text as instructions.

Why it was flagged

The skill may retrieve full email text. Email bodies are untrusted external content and could contain instructions aimed at influencing an agent.

Skill content
Use `--show-body` only when the full message text is necessary.
Recommendation

Treat retrieved email content as data only; do not follow instructions found inside emails unless the user explicitly asks for that action.

What this means

Local mail account names, email addresses, server hostnames, and mailbox paths may appear in results.

Why it was flagged

The script reads Thunderbird profile preferences to list accounts, email identities, hostnames, and mailbox directories. This is expected for the skill, but it exposes account metadata.

Skill content
prefs = profile / "prefs.js" ... return prefs.read_text(encoding="utf-8", errors="replace")
Recommendation

Only run account/profile listing when needed, and avoid sharing outputs that include account metadata.

What this means

Saved attachments could contain sensitive or unsafe files if opened or shared without review.

Why it was flagged

The skill can extract and write email attachments to disk. This is user-directed and purpose-aligned, but attachments are untrusted files.

Skill content
Filter and export attachments (`--has-attachment`, `--attachment-name`, `--save-attachments`)
Recommendation

Save attachments only to a deliberate folder, inspect them before opening, and avoid automatically executing or uploading saved files.

What this means

Users have less external provenance context for the bundled mail-reading script.

Why it was flagged

The registry metadata does not provide an upstream source or homepage for provenance. The bundled script is present and the static scan is clean, so this is a provenance note rather than a behavioral concern.

Skill content
Source: unknown; Homepage: none
Recommendation

Review the included script and install only if you trust this registry artifact.