plugin content safety

Content Safety Plugin

Trust-and-safety backend: evidence collection, legal holds, spam detection, raid protection, and abuse scoring. Pro plugin.

Requires: a Pro license tier or higher. Set it with nself license set nself_pro_... before installing.

Install

nself license set nself_pro_xxxxx...
nself plugin install content-safety

What It Does

Content Safety is the moderation and abuse-defense service for an nSelf backend, running on port 3213. It is the operator-facing trust-safety product layer — distinct from the always-free core hardening that ships with nself install, and distinct from nself-csam (PhotoDNA hashing).

• Trust-safety evidence: capture moderation evidence records, list them, and generate exports for review or external reporting.
• Legal holds: create and list legal holds that preserve evidence under a retention policy so it cannot be purged while a hold is active.
• Spam detection: analyze content against operator-defined spam rules, manage rule sets, configs, and per-action rate limits.
• Raid protection: record raid events, read and update raid status, and trigger lockdowns to defend against coordinated bursts.
• Abuse scoring: register and update per-actor trust/abuse scores that downstream apps can gate on.
• Multi-app isolation: every record is scoped by source_account_id (via X-Hasura-Source-Account-Id), so independent apps in one nSelf deploy never see each other's evidence, rules, or scores.

Configuration

Env Var	Default	Description
CONTENT_SAFETY_PLUGIN_PORT	3213	Service port
CONTENT_SAFETY_PLUGIN_HOST	0.0.0.0	Listen host
DATABASE_URL	—	Postgres connection string
CONTENT_SAFETY_API_KEY	—	Optional X-API-Key gate for the API

API

Health checks are at GET /health and GET /ready. All business endpoints live under /api/v1, grouped as evidence (/evidence, /evidence/exports), legal holds (/legal-holds), trust-safety stats (/trust-safety/stats), spam (/spam/analyze, /spam/config, /spam/rules, /spam/rate-limits), raid (/raid/status, /raid/lockdown), and abuse (/abuse/trust).

Licensing

This is a Pro, license-gated plugin (requires_license: true). Under the Security-Always-Free doctrine, core deployment hardening is free and automatic; Content Safety is an operator-facing moderation product (evidence retention, legal holds, configurable spam rules, raid lockdowns, abuse scoring), so it is licensed as a Pro feature.

See Also

• Plugin-Install — general plugin install flow
• Plugin-Catalog — full plugin list