Mod Dossier

One-tap user dossier for Reddit mods: in-subreddit history, removed-content samples, risk badges, shared mod notes, and inline ban/mute — right where you're already moderating.

Comment 1

Inspiration

Modding a busy subreddit on mobile is death by tap-tap-tap. Every borderline comment forces the same 30-second ritual: open the user's profile, scroll global history, mentally filter to "just this sub," switch contexts to a ban form, lose the original comment. Multiply by hundreds of decisions per day and you get mod burnout. Mod Dossier collapses that ritual into one tap and one screen.

What it does

Mod Dossier adds three moderator-only menu items to a subreddit:

  • Mod Dossier: open search on the subreddit overflow menu — type any username.
  • Mod Dossier: dossier author on any post's ⋮ menu — investigate the post's author.
  • Mod Dossier: dossier commenter on any comment's ⋮ menu — investigate the commenter.

Each opens a dossier post that shows, at a glance:

  • Account age, post karma, comment karma, verified email
  • Subreddit-scoped activity: posts, comments, posts removed, comments removed, removal rate %, approved count
  • Samples of the user's recent removed content in this sub
  • Risk badges: NEW ACCOUNT, LOW KARMA, MULTIPLE REMOVALS, HIGH REMOVAL RATE, VERIFIED EMAIL, TRUSTED MEMBER
  • Shared private mod notes (institutional memory for the whole mod team)
  • Inline Ban (reason + duration), Unban, Mute, Unmute

How I built it

  • Devvit Web 0.12.24 — HTTP server pattern with a webview front-end
  • TypeScript end to end, with a shared api.ts contract between client and server
  • Redis for shared mod notes (notes:{sub}:{user}) and dossier-target mapping
  • Reddit Plugin at moderator scope for user lookup, content history, ban/mute
  • esbuild for client and server bundling

The server computes the risk badges from rules tuned to surface the patterns mods actually care about: new throwaway accounts, repeat-offender removal patterns, and (positively) trusted long-time members so a mod can avoid over-reacting.

Challenges I ran into

  • Devvit Web's permissions block must explicitly declare redis: true and reddit: { scope: "moderator" } — otherwise the app deploys but the webview asset endpoint silently refuses connections.
  • Building three different entry points (subreddit, post, comment) that all converge on the same dossier UI, with the target user persisted per-post in Redis so the dossier survives page reloads.
  • Filtering Reddit's global user-history feed down to a single subreddit on the fly while keeping the response fast.

Accomplishments I'm proud of

  • Real moderation actions (ban / mute with reason + duration) work inline — no context switch
  • Shared mod notes give the mod team durable, queryable memory across shifts
  • Works on the Reddit mobile app, where most modding actually happens today

What I learned

  • The Devvit Web HTTP-server pattern is genuinely productive once permissions are correct
  • Risk thresholds need to be tunable per community — what counts as "high removal rate" varies wildly by sub
  • Subreddit-scoped signals beat global signals every time for moderation decisions

What's next

  • Per-subreddit configurable thresholds for risk badges
  • Optional flag-and-track of users automatically (cron + Redis history)
  • Cross-mod commentary on individual notes (replies / upvotes on notes)
  • An optional auto-comment on the original removed content linking to the dossier

Built With

Share this project:

Updates