Feature: Discord slash-command audit log visible in the WebUI admin panel

[lonix]

Summary

The WebUI already records every web-based state change in WebAuditLog (src/models/web-audit-log.ts, src/web/audit.ts). There is currently no equivalent visibility into Discord slash-command usage — operators cannot see who ran /quote, /config, /voicestats, etc., when they ran it, or whether it succeeded.

Adding a Discord-side command audit trail — and surfacing it in the existing admin WebUI — would close this gap with relatively little new infrastructure.

Proposed design

1. New Mongoose model: DiscordCommandAuditLog

// src/models/discord-command-audit-log.ts
{
  guildId:       String,  // required
  discordUserId: String,  // required
  commandName:   String,  // e.g. "quote", "config", "voicestats"
  subcommand:    String | null,
  channelId:     String,
  result:        "success" | "error" | "denied",
  errorMessage:  String | null,
  durationMs:    Number,  // wall-clock time from interaction received to reply
  createdAt:     Date,
}

2. Thin recording helper

Analogous to src/web/audit.ts:

// src/utils/record-command-audit.ts
export async function recordCommandAudit(entry: CommandAuditEntry): Promise<void> { ... }

3. Hook into CommandManager

src/services/command-manager.ts already has a central handleInteraction() dispatch path. Wrap each command execution with a try/catch that records success, error, or permission-denied outcomes.

4. New WebUI read-only page: /admin/audit/commands

• Table of recent command invocations (paginated, newest first)
• Columns: timestamp, user, command, result, duration
• Filters: date range, command name, user ID, result
• Reuse the existing AdminPageOptions / renderAdminPage layout

5. Retention / cleanup

Honour a new config key core.command_audit.retention_days (default: 90) in the existing VoiceChannelTruncationService cleanup cron, or a new dedicated cleanup job.

Value

• Operator visibility: identify abuse patterns (e.g. a user spamming /quote add), diagnose mysterious errors, verify that permission changes took effect.
• Security: correlate suspicious WebUI changes with prior slash-command activity.
• Low incremental cost: the model, helper, and WebUI page all follow patterns already established in the codebase (WebAuditLog, read-only-routes.ts, renderAdminPage). The most complex part is the CommandManager hook, which is a single try/catch wrapper.

Out of scope

• Recording raw command arguments (user privacy / potential secret leakage).
• Exporting audit rows via YAML (too large; use DB tooling for bulk export).