thekie/read-no-evil-mcp

AI Summary: This feature request proposes adding an `unread_only` parameter to the `list_emails` function. This parameter would allow users to filter emails to only retrieve those that are unread, leveraging the existing `is_seen` field. The primary motivation is to improve efficiency for AI agents by enabling server-side filtering of unread emails.

AI Summary: This issue proposes to improve error handling for the `rnoe.yaml` configuration file. It aims to replace raw Python tracebacks with user-friendly error messages for common mistakes like invalid YAML syntax, missing fields, incorrect ID formats, and file permission issues. Additionally, it suggests validating the configuration at startup to fail fast with clear messages.

AI Summary: This issue proposes adding a feature to notify agents when a newer version of the read-no-evil-mcp library is available. The notification will appear once per session on the first tool call if the agent is running an outdated version, prompting the user to update for the latest security protections. The implementation involves checking PyPI for the latest version and comparing it with the current version, with a fallback to silent failure if the check fails.

AI Summary: This issue proposes to preload the DeBERTa prompt injection model at server startup instead of lazy-loading it on the first request. This will eliminate unpredictable latency for the first user interaction and ensure that server readiness checks are accurate, though it will increase initial server startup time.

AI Summary: This issue proposes adding a `skip_protection: true` option to sender and subject rules to disable prompt injection scanning for specific senders. This allows users to bypass scanning for trusted internal systems or automated notifications, reducing latency and false positives, while maintaining scanning for other senders by default.

AI Summary: This issue proposes allowing email addresses as account IDs in the configuration. Currently, IDs are restricted to alphanumeric characters, but users often want to use their email addresses. The solution involves relaxing the ID validation to accept email-like strings and normalizing them by replacing non-alphanumeric characters with underscores and uppercasing for environment variable compatibility.

AI Summary: This issue highlights an inconsistency in YAML quoting styles within the project's documentation (README.md and CONFIGURATION.md). Some string values are quoted while others are not, despite both being valid YAML. The suggested fix is to choose a single quoting style and apply it uniformly across all examples for clarity.

AI Summary: This issue proposes enhancing the project's README with visual aids to improve understanding and engagement. The primary goal is to add an architecture diagram, with a terminal demo GIF as a secondary option, to better illustrate the system's flow and functionality.

AI Summary: This issue proposes adding a Frequently Asked Questions (FAQ) section to the project's README file. The goal is to address common user queries upfront, such as slow startup times, Gmail 2FA compatibility, and detection verification, thereby reducing support burden and improving user onboarding. The proposed topics cover a range of common concerns and point to existing resources where applicable.

AI Summary: This issue proposes adding a troubleshooting section to the project's README file. It aims to provide users with a self-service path for common problems by listing potential solutions for issues like IMAP connection failures, model download stalls, and permission errors. The proposed topics cover a range of common user challenges.

AI Summary: This issue proposes adding a new `examples/` directory to the repository to provide users with ready-to-use configuration files. These examples will cover various scenarios like Gmail, corporate IMAP, multi-account setups, and Docker Compose, each with detailed comments explaining the configuration fields. This aims to improve the user experience by eliminating the need to piece together configurations from README snippets.