Most requests never touch a model. A deterministic rule engine resolves clear-cut content in single-digit milliseconds, and only genuinely ambiguous content is escalated to AI. That keeps typical latency low enough to call CodifyNest inline, on the request path, without your users noticing.

One credit is one moderation call. A text moderation request costs one credit whether the rule engine resolves it or the AI layer is consulted, so your costs stay predictable. Batch requests charge one credit per item.

What happens if I run out of credits?

Your API keys keep working and the requests are billed as overage at your plan's rate, so moderation never silently fails open on you. You can set a hard cap in the console if you would rather have requests rejected than incur overage.

Do you store the content I send?

Only if you turn on moderation logs, which are off by default for content bodies. When logs are enabled you choose the retention window, and you can purge any record through the API or the console at any time.

Can I customise what gets blocked?

Yes. Policies let you set thresholds per category, add your own allow and deny terms, and decide the action for each outcome. Different policies can apply to different surfaces, so your DMs and your public forum do not have to share one standard.

Which languages are supported?

Text moderation covers 40+ languages, including transliterated and mixed-script text, which is where naive keyword filters usually fall over.

The API is plain HTTP and JSON, so it works from any language today. Official TypeScript and Python SDKs are on the way, and the docs carry copy-paste integrations for Discord, Slack, and Telegram bots.

How do I get started?

Create an API key in the console and send your first request. The free tier includes 10,000 credits a month and does not ask for a card, so you can evaluate CodifyNest against real traffic before you decide.

Content moderation that fits in one API call

CodifyNest classifies text, images, and prompts in milliseconds and returns a verdict you can act on. A deterministic rule engine handles the obvious cases; AI is called only when the content is genuinely unclear.

10,000 free credits every month · No credit card required

POST /v1/moderate/text

Illustrative response

Incoming content

hey does anyone know how to fix this build error? been stuck for an hour

Pipeline

Normalize…

Rule engine

Risk score

Verdict

200 OK · application/json

{
  "id": "mod_clean_2p9xKf",
  "action": "allow",
  "risk_score": 0.02,
  "resolved_by": "rules",
  "latency_ms": 6,
  "categories": [
    {
      "name": "harassment",
      "score": 0.01,
      "flagged": false
    },
    {
      "name": "hate",
      "score": 0,
      "flagged": false
    },
    {
      "name": "spam",
      "score": 0.03,
      "flagged": false
    },
    {
      "name": "sexual",
      "score": 0,
      "flagged": false
    }
  ],
  "policy": "community_default"
}

harassment0.01

hate0.00

spam0.03

sexual0.00

typical rule-engine latency: <10ms
languages supported: 40+
free credits every month: 10k

Products

One API. Every kind of unwanted content.

Start with text moderation today. The rest of the suite lands on the same base URL, the same auth, and the same response shape - so adding a product is never a re-integration.

Live

Text Moderation

Catch toxicity, harassment, hate speech, and threats before they reach your users.

POST /v1/moderate/text

Toxicity & hate
Harassment
Threats
40+ languages

Soon

Image Moderation

Scan uploads and avatars for NSFW, violent, and graphic content in a single call.

POST /v1/moderate/image

NSFW
Violence
Gore
URL or upload

Soon

Prompt Moderation

Defend LLM apps against jailbreaks, prompt injection, and adversarial input.

POST /v1/moderate/text

Jailbreaks
Injection
Data exfiltration
System-prompt leaks

Soon

PII Detection

Detect and redact emails, phone numbers, cards, and credentials before they leak.

POST /v1/moderate/text

Emails & phones
Cards & IBANs
API keys
Auto-redaction

Soon

Spam Detection

Block scams, phishing, and coordinated spam rings without punishing real users.

POST /v1/moderate/text

Scam links
Phishing
Raid patterns
Repeat offenders

How it works

Fast because most content never reaches a model

Sending every message to an LLM is the obvious design, and it is slow and expensive. CodifyNest resolves the clear-cut majority deterministically and spends AI only where it changes the answer.

Normalize

Unicode confusables, zero-width padding, leetspeak, and repeated characters are folded down. This is where naive keyword filters lose - i d i o t and 1d10t normalize to the same token.

Run rule engine

Deterministic matching against your policy: category thresholds, allow and deny terms, known scam domains, and repeat-offender patterns. No model, no network hop.

Calculate risk score

Signals collapse into a single 0-1 risk score per category, weighted by the policy you attached to the request.

Branch on confidence

Low risk returns allow. High, obvious risk returns block or review. Only the uncertain middle band costs anything more.

Escalate to AI - only if unclear

Ambiguous content goes to the model, which weighs intent, tone, and context. This is the expensive path, so the architecture is built to avoid it.

Return result

allowrisk < threshold

Ship it. No human ever sees it.

reviewunclear after AI

Queue it for a moderator with the reasoning attached.

blockrisk ≥ threshold

Reject it at the edge, before it reaches your users.

You decide what each verdict does. A Discord bot might map them to allow, delete, and timeout.

Integration

Live in about five minutes

No SDK required, no infrastructure to run, no model to fine-tune. Add one call to the path where content already enters your system.

Create an API key

Send your first request

One POST to /v1/moderate/text. Plain HTTP and JSON, so it works from any language you already use.

Act on the verdict

Branch on result.action. Map allow, review, and block to whatever your product already does.

curl -X POST https://api.codifynest.com/v1/moderate/text \
  -H "Authorization: Bearer $CODIFYNEST_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "You are such an idiot, get out of here",
    "policy": "community_default"
  }'

Response

200 OK · application/json

{
  "id": "mod_2p9xKfQ7Lm",
  "action": "block",
  "risk_score": 0.91,
  "resolved_by": "rules",
  "latency_ms": 8,
  "categories": [
    { "name": "harassment", "score": 0.94, "flagged": true },
    { "name": "hate",       "score": 0.11, "flagged": false },
    { "name": "spam",       "score": 0.02, "flagged": false }
  ],
  "policy": "community_default"
}

Use cases

Built for the places people actually talk

Bots, forums, comment threads, DMs, and chat UIs all have the same problem and the same shape of fix - a decision, made fast, on the request path.

Discord botsSlack appsTelegram botsForumsSocial appsAI chatbotsComment systemsLearning platforms

Discord bots

Moderate messages the moment they land, across every server you serve.

Allow · Delete · Timeout

Slack apps

Keep workspace channels clean without a human reading every message.

Allow · Flag · Remove

Telegram bots

Stop spam rings and scam links before they reach group chats.

Allow · Delete · Ban

Forums

Auto-triage new posts so moderators only see the genuinely unclear ones.

Publish · Queue · Reject

Social apps

Screen posts, comments, bios, and DMs through one consistent policy.

Allow · Review · Block

AI chatbots

Filter user prompts and model output on both sides of the conversation.

Allow · Rewrite · Refuse

Comment systems

Drop moderation into your existing comment flow with a single call.

Publish · Hold · Block

Learning platforms

Keep student submissions and discussions safe and age-appropriate.

Allow · Escalate · Block

Pricing

Pay for calls, not for seats

One credit is one moderation call, whether the rule engine answers it or the AI layer does. Start free, and move up only when your traffic actually does.

Messages moderated per day20,000

About 600,000 credits per month, at one credit per moderation call.

Recommended

Growth

$99/mo

Explorer

Everything you need to evaluate CodifyNest on a real workload.

Free

10,000

credits / month

10,000 credits / month
Text moderation
1 policy
Community support
No credit card required

Builder

For side projects and bots that outgrew the free tier.

$9/mo

100,000

credits / month

100,000 credits / month
Text moderation
5 policies
Webhooks
Email support

Startup

For production apps with real users and real moderation load.

$29/mo

500,000

credits / month

500,000 credits / month
All moderation products
Unlimited policies
Moderation logs & audit trail
Priority email support

Growth

For platforms moderating at scale across many communities.

$99/mo

2,000,000

credits / month

2,000,000 credits / month
All moderation products
Batch moderation
Team members & roles
99.9% uptime SLA

Enterprise

Volume pricing, custom models, and deployment on your terms.

Custom

credits - volume priced

Custom credit volume
Custom rules & models
Dedicated infrastructure
Custom SLA & DPA
Dedicated support channel

Contact sales

All plans include the full REST API, policies, and webhooks. Need volume pricing, a custom model, or a DPA? Talk to us.

FAQ

Questions, answered

If something here does not cover it, we would rather you asked than guessed.

Ship moderation this afternoon

10,000 free credits every month, no credit card, no sales call. Create a key and send your first request in the time it takes to read this page.

Talk to us

curl -X POST https://api.codifynest.com/v1/moderate/text \
  -H "Authorization: Bearer $CODIFYNEST_API_KEY" \
  -d '{ "text": "your first message" }'

Content moderation that fits in one API call

One API. Every kind of unwanted content.

Text Moderation

Image Moderation

Prompt Moderation

PII Detection

Spam Detection

Fast because most content never reaches a model

Normalize

Run rule engine

Calculate risk score

Branch on confidence

Escalate to AI - only if unclear

Live in about five minutes

Create an API key

Send your first request

Act on the verdict

Built for the places people actually talk

Discord bots

Slack apps

Telegram bots

Forums

Social apps

AI chatbots

Comment systems

Learning platforms

Pay for calls, not for seats

Explorer

Builder

Startup

Growth

Enterprise

Questions, answered

How fast is the API?

What is a credit?

What happens if I run out of credits?

Do you store the content I send?

Can I customise what gets blocked?

Which languages are supported?

Is there an SDK?

How do I get started?

Ship moderation this afternoon