LLMIFY THIS MOFO
CI / Test (push) Successful in 44s

This commit is contained in:
2026-09-01 01:49:40 -04:00
parent dad09d1109
commit 4426af63cb
19 changed files with 388 additions and 56 deletions
+21 -3
View File
@@ -47,6 +47,9 @@ func Save(cfg *Config) error {
for name, pc := range cfg.Providers {
sb.WriteString(fmt.Sprintf("[providers.%s]\n", name))
sb.WriteString(fmt.Sprintf(" api_key = %q\n", pc.APIKey))
if pc.BaseURL != "" {
sb.WriteString(fmt.Sprintf(" base_url = %q\n", pc.BaseURL))
}
sb.WriteString("\n")
}
}
@@ -117,6 +120,14 @@ func Save(cfg *Config) error {
sb.WriteString(fmt.Sprintf(" ydotool_timeout = %q\n", cfg.Injection.YdotoolTimeout.String()))
sb.WriteString(fmt.Sprintf(" wtype_timeout = %q\n", cfg.Injection.WtypeTimeout.String()))
sb.WriteString(fmt.Sprintf(" clipboard_timeout = %q\n", cfg.Injection.ClipboardTimeout.String()))
sb.WriteString(" ctrl_shift_v_classes = [")
for i, class := range cfg.Injection.CtrlShiftVClasses {
if i > 0 {
sb.WriteString(", ")
}
sb.WriteString(fmt.Sprintf("%q", class))
}
sb.WriteString("]\n")
sb.WriteString("\n")
// Notifications
@@ -221,6 +232,9 @@ keywords = []
# api_key = "" # ElevenLabs API key (or set ELEVENLABS_API_KEY env var)
# [providers.deepgram]
# api_key = "" # Deepgram API key (or set DEEPGRAM_API_KEY env var)
# [providers.llama-swap]
# api_key = "" # Or set LLAMA_SWAP_API_KEY
# base_url = "http://llama-swap.example:8080" # No /v1 suffix
# ─────────────────────────────────────────────────────────────────────────────
# Audio Recording
@@ -241,7 +255,7 @@ keywords = []
# ─────────────────────────────────────────────────────────────────────────────
[transcription]
provider = "openai" # "openai", "groq-transcription", "mistral-transcription", "elevenlabs", "whisper-cpp"
provider = "openai" # Also "llama-swap" for a remote OpenAI-compatible LlamaSwap server
model = "whisper-1" # Model: OpenAI="whisper-1", Groq="whisper-large-v3", Mistral="voxtral-mini-latest", ElevenLabs="scribe_v1"
language = "" # ISO 639-1 code (e.g., en, es, de). Empty for auto-detect.
threads = 0 # CPU threads for local transcription (0 = auto: uses NumCPU-1)
@@ -253,7 +267,7 @@ keywords = []
[llm]
enabled = true # Enable LLM post-processing (highly recommended)
provider = "openai" # "openai" or "groq" (must have API key configured above)
provider = "openai" # "openai", "groq", or "llama-swap" (must have API key configured above)
model = "gpt-4o-mini" # OpenAI: "gpt-4o-mini", Groq: "llama-3.3-70b-versatile"
[llm.post_processing]
@@ -272,7 +286,7 @@ keywords = []
# ─────────────────────────────────────────────────────────────────────────────
[injection]
backends = ["ydotool", "wtype", "clipboard"] # Ordered fallback chain (tries each until one succeeds)
backends = ["clipboard-paste", "ydotool", "wtype", "clipboard"] # Ordered fallback chain (tries each until one succeeds)
ydotool_timeout = "5s" # Timeout for ydotool commands
wtype_timeout = "5s" # Timeout for wtype commands
clipboard_timeout = "3s" # Timeout for clipboard operations
@@ -322,12 +336,16 @@ keywords = []
# - "groq-transcription": Groq Whisper API (very fast, models: whisper-large-v3, whisper-large-v3-turbo)
# - "mistral-transcription": Mistral Voxtral API (excellent for European languages, model: voxtral-mini-latest)
# - "elevenlabs": ElevenLabs Scribe API (99 languages, models: scribe_v1, scribe_v2, scribe_v2_realtime)
# - "llama-swap": Remote OpenAI-compatible LlamaSwap (any configured transcription model)
#
# LLM providers (for post-processing):
# - "openai": GPT models (gpt-4o-mini recommended for cost/quality balance)
# - "groq": Fast inference (llama-3.3-70b-versatile recommended)
# - "llama-swap": Any chat model configured in your LlamaSwap server
#
# Injection backends:
# - "clipboard-paste": Temporarily uses the regular clipboard + wtype Ctrl+V, then restores prior text clipboard.
# - ctrl_shift_v_classes: Window-class substrings that use Ctrl+Shift+V instead (default: ["ghostty"]).
# - "ydotool": Uses ydotool (requires ydotoold daemon). Best for Chromium/Electron apps.
# - "wtype": Uses wtype for Wayland. May have issues with some Chromium apps.
# - "clipboard": Copies to clipboard only (most reliable, requires manual paste).