Roadmap

Roadmap

General Information

• Privileges: This program requires ADMINISTRATOR privileges to function correctly.
• Forum and Thread Support:
  • Threads are treated as channels with additional considerations.
  • Operations like webhooks and slow mode require retrieving the parent channel.
• Emotion Scoring:
  • Major Scale: Ranges from 10 (love) to -10 (hatred), defining overall interaction tone.
  • Minor Scale: Adjusts per input (-1 to 1) to fine-tune tone incrementally.
  • Emotional resonance persists across channels for the same user.

---

Supported AI Engines (Completed)

• OpenAI
• Ollama
• Together.AI (Free models available)
• Cohere (Trial: 1000 free requests/month)
• HuggingFace (Free: 1000 requests/day; fragmented responses consume multiple requests)
• Anthropic (Pay upfront; token counting functional but complex)
• Perplexity.AI (Search-focused; excellent for lookup tasks)
• OpenRouter.AI:
  • Supports multiple providers, including paid and free models.
  • Basic OpenAI compatibility but may have limitations with normalization services.

---

Planned AI Engine Support (Exploration/In Progress)

• DeepInfra
• Fireworks.AI
• Anyscale
• Replicate
• Google.AI (Genini/Vertex AI):
  • API issues make development challenging.
• AI21 Labs:
  • Currently broken in key areas, not operational.

---

User Functionality

Completed

• %http: Reads URLs, YouTube transcripts, and PDFs.
• %yttags: Retrieves YouTube video tags.
• %Forget: Clears conversation memory in the current channel.
• %AnagramSolver: Solves anagrams.

In Progress/Planned

• Anti-nuke protection
• Ticket management system
• Image moderation: Includes anti-nudity verification.

---

Developer/Admin Features

Completed

• %PurgeRequests: Clears the server request queue.
• %CheckBot: Verifies if AI is allowed in a specific channel.
• Automatic Memory Purging: Removes memory older than X days.
• Impersonation Detection
• Auto Slow Mode: Mitigates spam.
• Anti-Raid Measures
• Personal Information Detection:
  • Identifies email, phone numbers, SSN, EIN, and credit card numbers.
  • Enhanced with AI classifiers.
• URL Verification: Uses AbuseIPDB.
• Content Identification/Moderation system:
  • Toxicity
  • Severe Toxicity
  • Identity Attack
  • Insult
  • Profanity
  • Threat
  • Sexually Explicit
  • Flirtation
  • Personal Attack
  • Inflammatory
  • Obscene