Skip to content

Releases: J3tze/ST-Tag-Pilot

v1.3.0 — Connection Manager Integration

19 Feb 11:57
@J3tze J3tze
v1.3.0
09ff188

Choose a tag to compare

View all tags
v1.3.0 — Connection Manager Integration Latest
Latest

Connection Manager Integration

Tag Pilot now uses SillyTavern's Connection Manager instead of storing API keys directly. This is a security improvement — your API keys are stored server-side in ST's secrets system, never in extension settings.

What's New

  • Connection Profile selector — select any of your existing Connection Manager profiles from a dropdown in Tag Pilot settings
  • No more API URL / Key / Model fields — configuration is handled by the profile
  • Secure API key storage — keys stay server-side in ST's secrets.json

Still Configurable

  • Temperature (default 0.4 for consistent tag formatting)
  • Max tokens (default 600)
  • Context depth, system prompt, user template
  • Per-character include/exclude tags
  • Skip quality prefix toggle

Requirements

  • SillyTavern 1.12.14 or newer (for Connection Manager support)

How to Update

Extensions → Update next to Tag Pilot → hard-refresh the page (Ctrl+Shift+R). After updating, select a connection profile in Tag Pilot settings.

Assets 2
Loading

v1.2.3

18 Feb 21:07
@J3tze J3tze
v1.2.3
fa7abda

Choose a tag to compare

View all tags
v1.2.3

Tag Pilot v1.2.3

New Features

  • SD command interception — Wraps the /sd slash command to generate tags via secondary LLM before image generation, bypassing the main LLM entirely
  • Wand dropdown interception — DOM capturing-phase listeners intercept SD dropdown button clicks before ST handles them
  • Mobile support — touchend listener with race condition fix ensures reliable interception on mobile devices
  • Unlock max tokens — Option to uncap the max tokens slider with a typable input field

Bug Fixes

  • Tag leakage cleanup — Uses MESSAGE_RECEIVED hook to strip raw tag strings from chat messages, preventing persistence after ST restart
  • Mobile double-fire — stopImmediatePropagation() called before guard check to prevent duplicate event handling

Other Changes

  • Updated recommended model to DeepSeek V3.2
  • Improved number input styling
  • Added Mermaid diagram to README
Loading

v1.2.1 — First Public Release

16 Feb 15:00
@J3tze J3tze
v1.2.1
02354a1

Choose a tag to compare

View all tags
v1.2.1 — First Public Release

Tag Pilot v1.2.1

Route image tag generation to a secondary LLM — cheaper, faster, uncensored.

Features

  • Auto-intercept — Automatically replaces SD prompts with booru-style tags via SD_PROMPT_PROCESSING event
  • Character-aware — Reads character card description for accurate appearance tags (hair, eyes, body, etc.)
  • POV-optimized — Default prompt template generates first-person POV tags for Pony/SDXL models
  • Post-processing — formatBooru() normalizes any model output into proper booru tags with quality prefixes (score_9, score_8_up, etc.) and auto-detected NSFW ratings
  • Per-character tag overrides — Always-include and always-exclude tags saved per character, auto-switching on chat change
  • Skip quality prefix — Toggle for use with SD style presets that provide their own quality tags
  • Slash commands — /gentags, /tags, /tagpilot for standalone tag generation, pipe-able into /sd
  • Graceful fallback — On any failure, the original SD prompt passes through unchanged
  • Image context filtering — Strips image-only messages from chat context sent to the LLM

Compatible Endpoints

Any OpenAI-compatible /v1/chat/completions endpoint: OpenRouter, DeepSeek, Ollama, LM Studio, vLLM, etc.

Install

Copy this URL into Extensions > Install Extension:

https://github.com/J3tze/ST-Tag-Pilot
Loading