Documentation / Strategy & Roadmap / Pending Implementations
⏳ Strategy & Roadmap

Pending Implementations

Tracked in-progress features, enhancements, and backlog.

Pending Implementations

Features with partial/stub implementations that need to be completed.


Info & System

/image <path> β€” Attach image for vision processing

Current state: Implemented.
Location: internal/core/kernel.go β€” Brain.AttachImage()

  • Integrate with LLM providers supporting vision (GPT-4o, Claude Sonnet, Gemini)
  • Read image bytes, encode base64 and pass in Chat request
  • Support common formats: PNG, JPEG, WebP
  • Size limit handling (≀20MB)

/paste β€” Attach clipboard image

Current state: Implemented.
Location: internal/core/kernel.go β€” Brain.PasteImage()

  • Read image from system clipboard (PowerShell on Windows, pngpaste on macOS, xclip on Linux)
  • Save clipboard to temp file and attach via AttachImage

/credits β€” Show credit/usage balance

Current state: Shows links to provider dashboards.
Location: internal/core/kernel.go β€” Brain.CreditsInfo()
Brownie points:

  • Query OpenRouter API for credit balance if provider is OpenRouter
  • Show cached balance from last API response headers (many providers return x-ratelimit-remaining)

/billing β€” Billing management

Current state: Shows links to provider billing dashboards.
Location: internal/core/kernel.go β€” Brain.BillingInfo()
Brownie points:

  • Same as /credits β€” provider-specific API queries

Configuration

/fast <mode> β€” Toggle fast mode

Current state: Implemented.
Location: internal/core/kernel.go β€” Brain.FastMode()

  • Store “fast model” override in Brain (gpt-4o-mini)
  • /fast on saves current model, switches to fast model
  • /fast off restores original model
  • /fast <model> allows custom fast model switch

/busy <mode> β€” Control Enter behavior while agent works

Current state: Implemented.
Location: internal/core/kernel.go β€” Brain.BusyMode()

  • Modes: queue (default β€” queues messages), steer (injects as steer), ignore (discards input while busy)

/voice <on|off> β€” Toggle voice mode

Current state: Not implemented.
Note: Requires TTS (text-to-speech) integration. Low priority for a programming-first agent.


Tools & Skills

/kanban β€” Project board from chat

Current state: Not implemented.
Note: Complex feature. Would need GitHub Issues integration or a local kanban data store.


Messaging (Gateway)

Commands from Hermes that would apply to gateway mode (Telegram, Discord, etc.):

  • /sethome β€” Set current chat as delivery home
  • /approve β€” Approve pending dangerous command
  • /deny β€” Deny pending dangerous command
  • /commands β€” Browse all commands (paginated)
  • /restart β€” Gracefully restart gateway
  • /update β€” Update GAIA to latest version
  • /topic β€” Multi-session DM mode (Telegram)

These require gateway integration and are lower priority than core features.