Free SKILL.md scraped from GitHub. Clone the repo or copy the file directly into your Claude Code skills directory.
npx versuz@latest install vivekkarmarkar-claude-code-os-skills-extract-tg-exchangegit clone https://github.com/VivekKarmarkar/claude-code-os.gitcp claude-code-os/SKILL.MD ~/.claude/skills/vivekkarmarkar-claude-code-os-skills-extract-tg-exchange/SKILL.md--- name: extract-tg-exchange description: Given vague or specific context about a Telegram exchange, find the matching entry in Telegram_calls.md and extract all available information (metadata + Claude's reply) --- # extract-tg-exchange — Find and extract a Telegram exchange ## Arguments The user provides context about a specific Telegram exchange. This context may be vague or specific: - `/extract-tg-exchange the one where I asked about what gets logged` - `/extract-tg-exchange message 2212` - `/extract-tg-exchange the rule 4 exchange` - `/extract-tg-exchange the one about the vibe physics article` - `/extract-tg-exchange the first voice message` - `/extract-tg-exchange save to output.md` (optionally save to a file) ## Behavior 1. **Read `Telegram_calls.md`** in the project root. This file is an audit log of Telegram reply tool invocations, with entries separated by `---`. Each entry typically contains: - Timestamp - Mode (audio/text) - Chat ID - Reply text (Claude's response) - Input prior to tool use (raw channel metadata) 2. **Match the user's context** to the correct entry. Use any available signals: - Message IDs mentioned in the metadata - Keywords in the reply text - Timestamps - Mode (audio vs text) - Content similarity to what the user described 3. **If ambiguous**, present the top candidates and ask the user to pick. 4. **Return all available information** for the matched exchange: - The full metadata block (raw channel tag) - Claude's reply text - Timestamp - Mode 5. **If the user requests saving to a file** (via "save to <file>", "put it in <file>", or similar), write the extracted exchange to that file. Use `/new-md` conventions (underscore naming, project root default). ## Output Format Present the extracted exchange clearly with labeled sections: - **Timestamp** - **Mode** - **Metadata** (the raw channel block) - **Claude's Reply Text**