Transcribe Telegram voice messages and forward to agents. Use when user sends voice/audio message via Telegram.
Transcribe Telegram voice messages and forward to agents. Use when user sends voice/audio message via Telegram.
npx skills add https://github.com/borodich/personal-os-skills --skill voice-telegram复制此命令并粘贴到 Claude Code 中以安装该技能
Set up your Personal OS in 15 minutes through a conversational interview. Creates SOUL.md, USER.md, IDENTITY.md, AGENTS.md, and MEMORY.md — the foundation files your agent reads every session.
Framework for building AI agents that evolve with their owner. Use when: setting up a new agent from scratch, onboarding a team to AI-native workflow, explaining the architecture to others, or auditing an existing agent setup for gaps.
Structured protocol for fixing bugs with AI agents. Prevents hallucinations and fix loops by enforcing step-by-step diagnosis before code changes.
Generate social media posts and long-form content in a consistent authorial voice. Use when writing posts for Telegram channels, blogs, or social media in a specific owner-defined style.
Prepare contact dossiers before meetings. Use when you need to research someone before a meeting, create a contact card in Obsidian, or check upcoming meetings that lack a briefing.
Load context from past sessions. Three modes: temporal (what did I work on yesterday/last week), topic (semantic search across sessions and notes), and graph (visual map of session-file relationships). Every recall ends with One Thing — the single highest-leverage next action.
| name | voice-telegram |
| description | Transcribe Telegram voice messages and forward to agents. Use when user sends voice/audio message via Telegram. |
Транскрибирует входящие войсы через faster-whisper и передаёт текст агенту как обычный запрос.
Когда приходит voice message из Telegram:
/Users/abserver/.openclaw/media/inbound/cd ~/Projects/voice-bridge && source .venv/bin/activate && python3 -c "
from stt.faster_whisper import FasterWhisperSTT
import sys, pathlib
path = sys.argv[1]
audio = pathlib.Path(path).read_bytes()
stt = FasterWhisperSTT(model_size='base', language='ru')
result = stt.transcribe(audio)
print(result or '')
" <PATH_TO_FILE>
Если транскрипт содержит "superflow" или "суперфлоу" — читай ~/clawd/skills/superflow/SKILL.md и запускай фазу 1.