read-no-evil-mcp

ReviewAudited by ClawScan on May 10, 2026.

Overview

The skill’s behavior matches its email-management purpose, but it can read, send, move, and delete email through a trusted MCP server, so users should configure permissions carefully.

Install only if you want an agent-accessible email gateway. Start with read-only accounts, enable send/delete/move only when necessary, use trusted local or secured remote MCP servers, protect the .env/config files, and verify the Docker image before giving it real mailbox credentials.

Findings (4)

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

If the configured MCP account allows it, the agent may send emails or change/delete messages.

Why it was flagged

The skill intentionally exposes email mutation capabilities, including sending, moving, and deleting messages. This is aligned with the advertised purpose, but those actions can have real mailbox impact if enabled.

Skill content
Use for reading, sending, deleting, and moving emails.
Recommendation

Keep accounts read-only unless write actions are needed, restrict send/delete/move permissions per account, and ask for confirmation before important sends or deletions.

What this means

The MCP server container can authenticate to the configured email accounts.

Why it was flagged

The optional Docker setup passes email account passwords from the user’s .env file into the MCP server container. This is expected for a mail gateway, but it means the server/container is trusted with mailbox credentials.

Skill content
if [[ "$line" == RNOE_ACCOUNT_*_PASSWORD=* ]]; then
        env_args+=("-e" "$line")
Recommendation

Use app-specific passwords or least-privilege mail credentials where possible, protect the .env file, and only run the server image if you trust it.

What this means

A compromised or changed container image could affect mailbox privacy or behavior.

Why it was flagged

The optional setup pulls and runs an external Docker image by tag. This is disclosed and purpose-aligned, but the image is not pinned by digest and will handle sensitive email access.

Skill content
IMAGE="ghcr.io/thekie/read-no-evil-mcp:0.3"
...
docker pull "$IMAGE"
...
docker run -d \
Recommendation

Verify the upstream project and image, consider pinning a known-good digest, and avoid using the Docker setup with sensitive accounts unless you trust the image.

What this means

If the remote server or network path is untrusted, email content and actions could be exposed or manipulated.

Why it was flagged

The skill can connect to a remote MCP server over an HTTP URL. Email commands and returned message content depend on that server’s trustworthiness and transport security.

Skill content
Remote server — An existing server on another machine. You need the URL (e.g. `http://server:8000`).
Recommendation

Prefer localhost for sensitive mail, or use only trusted remote servers with appropriate network protections such as HTTPS, authentication, VPN, or firewall restrictions.