Projects Directory
Bitsocial is meant to grow into thousands of apps, tools, and services built by anyone, not approved by gatekeepers. If you’re a dev, submit your own app.
OpenAI-compatible moderation checks against each community's rules.
AI Moderation Challenge evaluates Bitsocial comment content against community.rules with an OpenAI-compatible model endpoint. Communities can route risky posts to review while keeping provider keys and prompts in private node settings.
Custom image captchas for communities that want human verification.
Captcha Canvas Challenge generates visual captchas that communities can plug into their own publishing flow. It is a lightweight option for communities that want direct human checks without giving up self-hosted moderation.
Robot9000-style originality gate for communities that want anti-repost posting UX.
R9K Challenge runs on a Bitsocial community node and compares normalized post text against the owner node's local comments database. It gives communities a Robot9000-style posting experience without using AI: exact reposts fail before acceptance, backlinks are ignored, repeated failures trigger escalating temporary bans, and accepted-text hashes stay in private local state instead of storing raw post text.
Centralized risk scoring layer for filtering abusive publications.
Spam Blocker evaluates publications and returns a risk score that communities can combine with their own moderation logic. Its hosted service also links signer keys first seen from the same IP, so a user who keeps generating new accounts to dodge a community ban or manufacture a conversation sees their risk score rise, while a few signers stay allowed for legitimate multi-device use.
Community wordfilters enforced as a posting rule instead of a display filter.
Wordfilter Challenge makes a community's word replacements binding. Publishing clients apply the replacements before signing, so the signed comment and its CID already contain the replacement text, and the community node rejects any publication that still contains a filtered word. It implements the wordfilter/v1 contract, so a client written against that contract keeps working with other implementations.