A structured prior for reasoning about whether an app is appropriate for a minor. Every app gets two independent risk scores — safety and attention — plus one derived disposition, adjusted by four age bands. Apps not listed are resolved through fallback heuristics. It is a default, not a verdict: the specific child, family rules, and observed use always outweigh the prior.
The disposition scale — five verdicts absorbing the informal “appropriate / distraction / watch / no-no” tiers. Derived from the two risk axes plus the child’s age band. Click a level to jump to its app group.
Dense reference, not a script — the prior brought to a decision before looking at the specific child, family rules, and actual usage.
If listed, its disposition + risk axes + age-band notes are the starting point.
Derive the same signals from category, age rating, and features via the unknown-app heuristics.
A browser doing homework and a browser in a YouTube rabbit hole are the same app with opposite dispositions. Weight age, family values, and observed usage over the prior.
The hard-block list is categorical — illegal for minors, adult content, gambling — not a matter of degree.
A single label can’t capture that YouTube is a moderate safety risk and a high attention risk — different concerns needing different parental responses.
Exposure to harm — four concerns, weighted in this order of severity:
How hard the app is engineered to be habit-forming — Clearspace’s core concern. Signals:
↓ Safety × Attention × Age band → Disposition
Where representative apps land on the two axes. Chip color = disposition — note it is not a diagonal: high-safety-risk cells hold Block, Restrict, and Supervise apps side by side, because age gates and categorical concerns (gambling, adult content) move the verdict independently of position.
Rows read bottom-to-top: none → low → moderate → high safety risk. Apps with ranged scores (e.g. Snapchat’s M–H safety) are placed at their primary level in the full tables below. Hover a chip for the one-line rationale.
Guidance splits at the ages where consensus actually diverges. Column widths are proportional to the years each band spans.
No independent social media or open-comms; walled-garden / kids’ modes only.
Below the 13+ line for nearly all social platforms; presence on them is a flag.
Meets nominal age gates; the highest-supervision social window.
More autonomy; some “Restrict” apps become “Supervise.”
Age gates are the floor, not the ceiling: COPPA restricts data collection under 13; most social platforms require 13+; dating and gambling require 18+ (21+ for some betting).
The conventional wisdom the model encodes — numbered as in the source so the agent can cite them.
A child under a platform’s stated age is a categorical flag — they had to lie about their birthday to be there.
An app where unknown adults can reach a child is riskier than an app with merely mature content. Prioritize this.
YouTube, TikTok, Reddit, and X are as safe or as harmful as their algorithm serves. Reason about the feed and settings, not just the logo.
Loot boxes, gacha pulls, simulated casino, and pay-to-win chests normalize gambling and carry real spending — even in “games.”
Disappearing messages and stories make bullying, sexting, and stranger contact harder to detect. Raise supervision accordingly.
Real-time video + chat (Twitch, TikTok Live, Instagram Live) can’t be pre-screened. Higher risk than recorded content.
Browsers, YouTube, and messaging are dual-use. Judge observed behavior over the app’s mere presence; time-of-day and duration are strong signals.
Default posture is resolved via the fallback heuristics — not an automatic block and not an automatic pass.
Most productivity and developer tools are adult-work context. Don’t flag GitHub or a terminal as a child concern.
Risk columns: S = safety, A = attention, on the none → low → moderate → high scale. Ranged values (e.g. M–H) show a lighter final segment.
Categorical: age-gated to 18+/21+, adult content, or gambling. Never negotiable — the concern is not a matter of degree.
| App | Type | S | A | Why |
|---|---|---|---|---|
| Pornhub, xHamster | Adult content | H | H | Pornography. No minor use case. |
| OnlyFans | Adult content / creator | H | M | Predominantly adult content; 18+. |
| Tinder, Bumble, Hinge | Dating | H | M | 18+ dating; adult stranger contact, location. Even 16–17: block. |
| DraftKings, FanDuel, ESPN BET / theScore Bet | Sports betting / casino | H | H | Real-money gambling; 21+ in most US states. |
| Robinhood, Coinbase | Trading / crypto | H | M | Real-money speculative trading; minors can’t hold accounts. Gambling-adjacent. |
Nominal age gate is often 13+, but the unfiltered adult-content surface and/or stranger contact makes these inappropriate for younger minors. 16–17 may move to Supervise with privacy settings locked down.
| App | Type | S | A | Notes |
|---|---|---|---|---|
| X (Twitter) | Social | H | H | Unfiltered adult content, graphic media, misinformation; weak moderation. Block <16; Supervise 16–17. |
| Social / forums | H | H | Enormous NSFW surface; SFW by default but one tap away. Block <13; Restrict 13–15; Supervise 16–17. | |
| Tumblr | Social / blogging | M | M | Historically heavy adult & self-harm content. Restrict <16. |
| Telegram | Messaging | H | M | Minimal moderation; adult/extremist channels; stranger contact. Restrict; Supervise 16–17 for family/friends only. |
| Twitch | Livestream | M | H | Live video + unmoderated chat, mature streams. Restrict <13; Supervise 13+. |
The core “watch” tier. Fine for teens who meet the age gate if privacy settings are locked down, comms are limited to known contacts, and a parent stays in the loop. Almost always Block under 13, because the platform itself requires 13+.
| App | Type | S | A | Age & supervision notes |
|---|---|---|---|---|
| YouTube | Video | M | H | 13+ for main app. Under 10 → YouTube Kids only. 10–12 → supervised mode. Autoplay + recommendations drive attention risk; restricted mode filters some content. |
| Social | M | H | 13+. Block <13. Teen Accounts / private profile, no DMs from non-followers, limit Reels. Body-image content is the main content risk. | |
| TikTok | Social / video | M | H | 13+. Block <13. Highest attention risk of any app here (algorithmic short-form). Family Pairing + restricted mode. |
| Snapchat | Messaging / social | M–H | M | 13+. Block <13. Ephemerality lowers accountability; Snap Map = location; stranger contact via Quick Add. Lock down location & contacts. |
| Discord | Chat / communities | H | M | 13+. Block <13. Public servers + DMs = real stranger-contact risk. Supervise closely: known servers only, DMs off from non-friends. |
| Roblox | Gaming (UGC) | H | H | Massive under-13 base but significant predator/UGC/spending concerns. Supervise all ages: account restrictions, chat off/limited, spending controls. |
| Fortnite | Gaming | M | H | Rated T (13+). Voice chat with strangers + spending (V-Bucks). Disable open voice, cabined accounts for younger. |
| Steam | Game store / social | M–H | M | Storefront includes mature/adult games; chat with strangers. Family View + purchase controls. Supervise <16. |
| Messaging | L–M | L | 13+. Mainly known-contact messaging; group invites are the contact vector. Supervise light for teens. | |
| Facebook Messenger | Messaging | M | M | Contact via friend network; Messenger Kids is the walled-garden alternative for younger. |
| Social | M | M | Older-skewing; 13+. Marketplace & Groups are the stranger-contact surface. | |
| Social / discovery | L–M | M | 13+. Mostly benign; historical diet/self-harm content is the residual concern. Supervise light. | |
| Threads | Social | M | M | Tied to Instagram account; 13+. Similar posture to Instagram, less algorithmic pull today. |
| Locket, Partiful, BeReal-style, Alfi | Social / photo | M | L | Photo/location sharing with a contact circle. Watch who’s in the circle. |
Low safety risk; the concern is time, not harm. Cap duration, prefer kid profiles / explicit-content filters for younger children, don’t forbid.
| App | Type | S | A | Notes |
|---|---|---|---|---|
| Netflix, Disney+, Hulu, Prime Video, YouTube TV | Streaming video | L | M | Use Kids profiles + maturity PIN for under-13. Binge/autoplay is the attention cost. |
| Spotify, Apple Music, YouTube Music, SoundCloud | Music | L | L–M | Enable explicit-content filter for younger. Podcasts within can be mature. |
| Audible, Apple Podcasts, Overcast | Audio / podcasts | L | L | Low risk; content maturity varies. |
| Minecraft | Gaming (creative) | L–M | M | Creative/educational core; online multiplayer servers add contact risk → supervise online play. Offline/realms with friends = Allow. |
| Clash Royale, Clash of Clans, Brawl Stars | Gaming | M | M | Loot-box/chest spending mechanics + clan chat. Watch in-app purchases; otherwise moderate. |
| Block Blast, casual/puzzle, NYT Games, Bandle | Gaming (casual) | L | M | Benign; ad-supported free ones may push purchases. |
| CapCut | Video editing | L | M | Creative tool; tied to TikTok ecosystem. |
| Amazon, Zara, retail apps | Shopping | L | L | Spending is the only real concern; disable 1-click / require approval. |
| Apple News, NYT, AP News, Google News | News | L–M | M | Occasional mature content; low kid relevance. |
Low on both axes. Green-light.
| App | Type | S | A | Notes |
|---|---|---|---|---|
| Duolingo | Education | — | L | Language learning; streaks are a mild attention hook, not a risk. |
| Khan Academy, Minecraft Education, Quizlet, Photomath, Prodigy | Education | — | L | Core learning tools. |
| Chess.com, Lichess, Chess | Games (strategy) | L | L | Benign; chat features exist but are minor. Chess.com is 13+ nominally. |
| YouTube Kids | Video (walled) | L | L | The under-10 substitute for YouTube. Still cap time. |
| Photos, Calculator, Weather, Maps, system utilities | Utility | — | — | No concern. |
When an app isn’t listed, derive its disposition from these questions — in priority order. When genuinely uncertain, default to the more cautious of two adjacent dispositions and surface why, so a parent can override with context the agent doesn’t have.