Sanctum

Criticism stays.
Abuse gets hidden.

Your comments, on your terms. Sanctum hides high-confidence harmful comments before they reach you. Hidden, not deleted. Reversible any time.

One email when Sanctum opens to new accounts. Nothing else, ever.

See how it works ↓

How it works

Four steps. Set once.

Sanctum sits between Instagram and you. Nothing to configure beyond picking a protection mode.

  1. 01

    Connect

    Link your account once. Sanctum gets only the moderation permissions it needs — no DMs, no insights, no posting.

  2. 02

    Watch

    Every new comment is checked in real time. Threats, hate, personal attacks, spam, passive-aggression — each category has its own confidence threshold.

  3. 03

    Hide

    High-confidence harmful comments are hidden — never deleted. You never see them. Disagreement, questions, and criticism stay visible.

  4. 04

    Audit

    Open the audit log any time. Review what Sanctum hid. Restore anything you disagree with. Per-comment, always reversible.

The line

What gets hidden.
What stays visible.

Sanctum has a strict line — and you can see exactly where it is. No guessing, no shadow rules.

Sanctum hides

  • Threats and incitement to harm The clearest line — hidden in every mode.
  • Hate speech and slurs Identity-based attacks. High confidence required.
  • Personal attacks "You're worthless." "Kill yourself." Hidden.
  • Spam and scams Repeat scripts, sketchy links, follow-bots.
  • Passive-aggressive sneers "Must be nice." "Some of us…" Optional — depends on your mode.

Sanctum leaves alone

  • Honest disagreement "I disagree with this take" stays.
  • Questions and curiosity "Where did you find this?" stays.
  • Constructive criticism "Here's what I'd push back on" stays.
  • Corrections and clarifications "Small correction…" stays.
  • Jokes and sarcasm without a personal attack Banter stays.

Threshold values live in app/moderation/policy.py. Change them only as a deliberate product decision — not as tuning.

Trust

Three non-negotiables.

Not a startup roadmap — product principles that won't change. If Sanctum ever breaks one of these, it stops being Sanctum.

  1. Hide, never delete

    Sanctum can hide comments. It cannot delete them. The Instagram API allows both — we deliberately use only hide, because hiding is reversible. Every decision is recoverable.

  2. You never have to read it

    The dashboard shows counts, not the comment text. The audit log is opt-in, friction-gated, per-comment reveal. Sanctum does not normalise letting harassment into your day.

  3. Your data, your control

    Comment text is retained for 60 days, then purged. One-click "delete my data" wipes everything Sanctum holds about you. No sharing, no selling, no analysis beyond moderation.

Privacy Terms Delete my data

FAQ

Common questions.

What does Sanctum do?
Sanctum is an automated comment moderator for Instagram creators. It uses an LLM classifier to detect high-confidence harmful comments — threats, hate, personal attacks, spam, and passive-aggressive sneers — and hides them before you see them. It never deletes. Every decision is reversible.
How is it different from Instagram's built-in Hidden Words?
Hidden Words filters comments containing specific words you list. Sanctum reads context — it can hide a personal attack that contains none of your filter words, and it leaves alone comments that happen to use a "bad word" in honest context. The dashboard tells you how many it caught and which categories.
Does it handle spam and scam comments too?
Yes — spam and scams are one of the categories Sanctum hides, alongside threats, hate, personal attacks and passive-aggressive sneers. Bots, follow-for-follow scripts and sketchy links turn up on every post, not only the ones that attract abuse, so Sanctum is clearing your comments on a quiet week as well as a bad one. Same rules as everything else: hidden, not deleted, and listed in the audit log.
Will Sanctum hide legitimate criticism?
No. Disagreement, questions, corrections, and constructive criticism are categorised as neutral and never hidden, regardless of confidence. Only categories the policy lists as harmful — and only above their per-category threshold — get hidden.
Can I see what Sanctum hid? Can I restore it?
Yes to both. The audit log shows every hide with category, confidence, and date. The comment text is collapsed by default; revealing it requires a per-comment opt-in (no bulk text exposure). Any hidden comment can be restored — Sanctum calls Instagram's API to make it public again.
Does Sanctum store the comment text?
Hidden comment text is stored for 60 days so the audit log has something to show, then automatically purged. The hide-decision metadata (category, confidence, timestamp) is retained longer for dashboard counts and per-account adaptive moderation. Full details on the Privacy page.
When can I start using it?
Sanctum is invite-only right now and opening in stages, so every new account gets watched properly while its moderation settles in. Join the waitlist and you'll get one email when it's your turn — no drip campaign, no newsletter.
What does it cost?
One subscription tier with every protection feature included — no plan that withholds the thing you came for. The price is not final yet; the waitlist hears it first.

Who builds this

I'm Bartek. Solo developer in Norway — and Sanctum's first user.

On Instagram I'm @w_poszukiwaniu_troli ("Norwegia bez banału"). The comments started getting bad enough that I went looking for tools. Nothing on the market just quietly hid the worst stuff without turning the account into a fortress.

So I built one. Calm, firm, protective. Hide, never delete. Audit any time. Built first for me — now opening to a small group of tester creators.