Replace per-case trackers with a unified key-based tracker and tighten detection logic. Introduces buildMessageTrackingKeys/collectTrackableText to canonicalize text, extract invite codes, include image fingerprints and dedupe keys; replaces trackCrossChannelMessage/trackImageOnlySpam with trackCrossChannelKeys that tracks multiple normalized keys. Simplifies detectSpam to use trackable content and invite-aware link handling, updates message preview/attachment embedding, and adjusts command/help text to reflect the new "same content across channels" rule. Overall reduces duplicate-prior merging and streamlines cross-channel spam handling.
Add support for detecting and tracking image-only spam across channels. spamDetector.js: introduce image helpers (isImageAttachment, getImageAttachments), buildImageFingerprint, isImageOnlyMessage, and extend detectSpam to report image-only reasons and a duplicateImageAcrossChannels flag. rapidSpamTracker.js: refactor cross-channel tracking with trackCrossChannelPost, pruneExpiredPosts, dedupePriorMessages, and add trackImageOnlySpam which uses optional image fingerprints; export dedupePriorMessages. index.js: wire image-only tracking into message handling (build fingerprint, trackImageOnlySpam, pass duplicateImageAcrossChannels to detectSpam), merge and dedupe prior message lists, and include attachments in mod log previews. commands.js: update detection rules text to mention image spam. These changes enable identifying image-only spam (including fingerprinting) and avoid duplicate prior-message entries when creating moderation logs.
Introduce a rapidSpamTracker to detect same messages posted across channels within a short window and return prior message IDs for cleanup. Refactor spam detection to collect scannable text from message content, embeds and attachments, simplify role-ping and link heuristics, and mark spam when a role ping + link is duplicated across channels. Update index to use the tracker/collector, delete prior spam messages, improve moderation flow (ban → kick fallback), report clearer action/error labels, and warn about missing channel permissions. Update commands display text for detection/actions and bump .env.example MIN_ACCOUNT_AGE_DAYS default to 7.
Introduce a new Anti-Spam Discord bot project. Adds project metadata and scripts (package.json, .env.example, .gitignore, README, deploy-commands.js) and the full src implementation: command registration and handling, per-guild settings persistence, configuration parsing, spam detection logic (phishing URLs, suspicious TLDs, mass/role mentions, image+URL combos, new account heuristics), and the bot entrypoint (index.js) that enforces permissions, logs actions, and optionally bans offenders. README explains setup, required intents, and usage; per-server settings are stored under data/guild-settings.json. (package-lock.json added for dependency resolution.)
andrew
Andrew Zambazos